# Orchid: Full Context

> Orchid is the executive assistant that doesn't sleep. It works for founders, investors, lawyers, real estate agents, agency owners, and operators. It handles inbox triage, calendar coordination, meeting prep, travel, and follow-ups. Drafts stay drafts until you approve. Lives in Gmail, Google Calendar, and SMS. No new app to babysit.

## What Orchid does

Orchid reads your inbox, learns how you write, and starts handling the routine work in your voice. The investor who wants 20 minutes before they commit. The candidate who needs to reschedule. The vendor who has gone dark on the invoice. Orchid drafts the reply, surfaces the decision, and waits for your one-click approval.

Your morning starts with a brief: the threads that need you, separated from the noise that does not. The work that compounds stays in front of you. The rest moves in the background.

## How Orchid works

- **Reads everything**: your inbox, calendar, contacts, and connected tools.
- **Learns your voice**: tone, phrasing, how you write to a client versus a friend.
- **Prepares actions**: drafts replies, schedules meetings, preps you for calls, chases follow-ups.
- **Asks before sending**: nothing important goes out without your approval.
- **Surfaces only what needs you**: tier-1 notifications (OTPs, payment failures, magic links) always reach you; everything else respects your schedule.

## Frequently asked questions

### What is Orchid?

Orchid is your personal assistant, the kind of help that used to be reserved for a lucky few. It quietly handles your email, your calendar, your meeting prep, and your travel, so you can focus on what actually matters to you.

### What does Orchid do for me?

Orchid runs your inbox, books your meetings, and preps you before every important call. It plans your trips, sources thoughtful gifts, and handles the small stuff that quietly eats up your day.

### Who is Orchid for?

Orchid is built for people who care about their time, like founders, investors, lawyers, agents, and agency owners. It is for anyone whose hours should go toward real work, not the endless busywork around it.

### How much can Orchid handle?

Probably more than you would expect. Orchid handles email, scheduling, vendors, invoices, travel, meeting prep, and even gift sourcing. It can also book the doctor visit you keep putting off. You decide the scope, and Orchid takes care of the rest from there.

### How does Orchid learn my voice?

Orchid learns by reading what you have already written. It picks up your tone, the phrases you use, and how you write to a client versus a friend. The longer you use it, the more your drafts begin to sound exactly like you wrote them.

### Does Orchid ever act without me?

Only within the limits that you set. Replies stay as drafts until you hit send, and calendar holds happen only with your approval. You can pull back access at any time, and nothing important goes out without you.

### How does privacy work?

Orchid only sees the accounts that you choose to connect. We never sell your data, share it with anyone, or use it to train any model. You can disconnect or delete everything from settings whenever you want to.

### What apps does Orchid work with?

Today, Orchid works with Gmail and Google Calendar. Many more apps are on the way, like Outlook, Slack, Notion, and Linear. If you need one we have not added yet, tell us and we will move it up the list.

## Comparisons

## Orchid vs Poke

URL: https://orchid.ai/compare/orchid-vs-poke
Published: 2026-08-21
Facts verified: 2026-08-21 (source: https://poke.com/pricing)

Poke and Orchid share a surface: an assistant that lives in your messages. The difference is the job. Poke centers on channels and custom Recipe automations. Orchid is built as an executive assistant: full inbox triage with drafts held for your approval, meeting prep briefs from your email, calendar, and notes, a morning digest, and follow-ups that happen without you asking. If the point is getting the work handled with your approval on everything that leaves, that is Orchid.

### Where it lives

| Feature | Orchid | Poke |
| --- | --- | --- |
| Primary surface | iMessage and SMS | iMessage, WhatsApp, Telegram |
| Free tier | — | Yes |
| Proactive messages | Yes | Yes |

### Who does the work

| Feature | Orchid | Poke |
| --- | --- | --- |
| Email handling | Full triage, drafts held for your approval | Search, draft, send, organize labels |
| Outbound waits for your approval by default | Yes | — |
| Meeting prep briefs | Yes | — |
| Morning digest of what needs you | Yes | — |
| Reminders and scheduled tasks | Yes | Yes |

### Apps and automations

| Feature | Orchid | Poke |
| --- | --- | --- |
| Integrations | Gmail, Google Calendar, Granola, Slack, and more | 30+ apps |
| Custom automations | Workflows you set up by asking | Poke Recipes |
| Pricing model | Flat monthly | Flat, plus pay-as-you-go on Ultra |

### Pricing

- Poke Free: $0
- Poke Pro: $19/mo
- Poke Ultra: $199/mo plus usage
- Orchid: From $29.99/mo

## Same surface, different job

Poke and Orchid both live in your messages: no new app, no dashboard, an assistant as a contact in your texts. If you are choosing between them, you have already made the important decision.

From there the products diverge. Poke is organized around channels and automations: iMessage, WhatsApp, and Telegram, thirty-plus integrations, and Recipes for building your own flows, with a free tier and pay-as-you-go usage on its top plan.

Orchid is organized around one job: being your executive assistant. Every draft waits for your approval. Meeting prep briefs arrive before the call with context from your email, calendar, and notes. Your morning starts with a digest that separates the things that need you from everything that does not. Follow-ups happen on threads you forgot about, without you asking.

## Where Orchid is different

The approval loop is the core: nothing leaves without you, which matters once an assistant touches real correspondence. The context runs deeper: briefs before meetings, judgment about what deserves your attention, and follow-through that starts on its own. Orchid is priced and built like help you hire, not an app you try.

If the goal is handing over the work around your work, that is the job Orchid was built for.


### FAQ

### What is the real difference between Orchid and Poke?

Depth on the executive assistant job. Both live in your messages and both are proactive. Orchid holds every draft for your approval, briefs you before meetings with context from email, calendar, and notes, and opens your day with a digest of what actually needs you. Poke leans toward breadth: more channels, a free tier, and custom Recipe automations.

### How does pricing compare between Orchid and Poke?

Poke has a free tier, Pro at $19 per month, and Ultra at $199 per month with pay-as-you-go usage beyond its credits. Orchid starts at $29.99 per month, flat. Poke is priced to explore; Orchid is priced around replacing hours of assistant work.

### Do Orchid and Poke both work in iMessage?

Yes. Poke works through Apple Messages, WhatsApp, and Telegram. Orchid works through iMessage and SMS. Neither requires installing an app.

### Does Poke's Cognition acquisition change how it compares with Orchid?

Cognition acquired The Interaction Company, Poke's maker, in mid 2026, and said Poke continues operating for its users. Its team reported over 100 million messages exchanged in the three months before the acquisition.

### Do Orchid and Poke send email without your approval?

Orchid never does: every draft waits for your approval by design, so nothing leaves in your name without you seeing it. Poke's site describes searching, drafting, sending, and organizing email; an approval-first default is not something it advertises, so check its settings before connecting a real inbox.

## Writing

## Orchid on iMessage: What the AI Assistant Does and How to Set It Up

URL: https://orchid.ai/blog/orchid-ai-assistant-imessage
Published: 2026-09-14
Topic: Guide

Orchid lives in your texts. No app, no new inbox. What it actually does with your email and calendar, why we put it in Messages instead of an app, and how to set it up in two minutes.

Orchid lives in your texts. There's no app to download and no new inbox to check. You text a number, it texts back, and behind that thread it's reading your email, watching your calendar, and doing the work you'd otherwise hand to a person.

A lot of people find us by searching "Orchid AI assistant iMessage," and until now the best answer we had was the homepage. This is the longer one: what Orchid actually does from a text thread, why we put it there instead of in an app, and how to set it up. The setup part takes about two minutes.

## Why a text thread and not another app

Every app you install is another place work can hide. We wrote about that at length in [The Next App Is No App](/blog/the-next-app-is-no-app), but the short version is that adding one more destination never reduced anyone's coordination work. It added to it.

The research on this is blunt. Gloria Mark and her colleagues found that after an interruption, people took an average of 23 minutes to get back to what they were doing, and that they finished the interrupted work faster but with significantly more stress ([Mark, Gudith and Klocke, CHI 2008](https://www.ics.uci.edu/~gmark/chi08-mark.pdf)). Sophie Leroy showed that part of your attention stays behind on the last task even after you switch, which she called attention residue, and it makes you worse at the next one ([Leroy, 2009](https://doi.org/10.1016/j.obhdp.2009.04.002)). Microsoft's Work Trend Index put a number on where the day goes: 57% of it in communication tools, 43% actually creating, and 68% of people saying they don't get enough uninterrupted time to focus ([Microsoft, 2023](https://www.microsoft.com/en-us/worklab/work-trend-index/will-ai-fix-work)).

So we made a decision early. The assistant should come to the one thread you already have open, and it should never ask you to open anything else. For most people that thread is Messages. It's on your lock screen, it already has your attention, and it's already where you tell people what you need.

## What it does from the thread

Everything below happens in the same conversation. You don't switch modes and you don't open a dashboard.

Connect Gmail and Orchid starts reading what comes in. It surfaces the threads that need you and drafts the replies in your voice, and the drafts stay drafts until you say send. We wrote about that shift, from processing mail to delegating it, in [The End of Inbox Zero](/blog/the-end-of-inbox-zero).

Scheduling is a sentence. _find 30 minutes with sarah next week_ becomes a hold on your calendar and a note to Sarah. Reschedules, conflicts, and travel time get the same treatment, and nothing lands on the calendar without your okay.

Before a call, Orchid texts you who you're meeting, what you talked about last time, and what's still open between you. If you use Granola, [it carries your notes forward](/blog/orchid-and-granola) into the follow-ups.

Reminders work the way you'd say them to a person. _remind me to chase the landlord friday if he hasn't replied._ Orchid watches the thread and only nudges you if the condition is still true.

If you'd rather talk than type, [send a voice note](/blog/introducing-voice-notes). Orchid pulls out the actions and confirms what it heard. And when an answer doesn't fit in a bubble, an itinerary, a comparison, a plan, it comes back as [a stack of cards you swipe through](/blog/introducing-carousels).

None of this stops at work. People use the same thread to [log meals and workouts](/blog/i-lost-15-pounds-by-texting-an-ai), to [stay close to the people they keep meaning to call](/blog/the-people-you-mean-to-call), and to [get through a semester](/blog/how-i-used-ai-to-get-a-4-0).

## Setting it up

1. Text [+1 (415) 605-3216](sms:+14156053216). Say hi, or just say what you need.
2. Orchid sends you a link to connect Gmail and Google Calendar. It's Google sign-in, so you never hand Orchid a password, and you can revoke access from your Google account whenever you like.
3. Tell it how you work. Which senders matter, when you want the morning rundown, what it should never touch. It remembers.

That's it. Orchid learns your voice from what you've already written, so the first drafts are usable and they get better from there.

## Is it really iMessage?

On an iPhone, yes. The thread with Orchid is a regular blue-bubble conversation, with the same end-to-end encryption Apple applies to every iMessage ([Apple Platform Security](https://support.apple.com/guide/security/imessage-security-overview-secd9764312f/web)).

## What it costs, and what it won't do

Orchid is paid, from $29.99/mo. There's no free tier, because an assistant that reads your inbox shouldn't be paid for with your data. We lay out the reasoning, and the student and founder deals, in [Is Orchid AI free?](/blog/is-orchid-ai-free-pricing-explained)

It also won't send an email, book a meeting, or touch anything outside your thread without your approval. It only sees the accounts you connect, and we never sell your data or train on it. The long version is in [Is Orchid AI safe?](/blog/orchid-ai-reviews-reddit-is-it-safe) and the [privacy policy](/legal/privacy).

## Frequently asked questions

**Do I need to download an app?**

No. Orchid runs inside Messages. There's a web app at [app.orchid.ai](https://app.orchid.ai) for connecting accounts and changing settings, but you don't need it day to day.

**Which accounts can I connect?**

Gmail and Google Calendar today, through Google sign-in. Outlook, Slack, Notion, and Linear are on the way. Tell Orchid which one you need and it moves up the list.

**Can Orchid send emails on its own?**

Only within the limits you set. By default replies stay as drafts until you approve them, and calendar holds wait for your sign-off.

**How is this different from asking ChatGPT in Messages?**

A chat model answers questions. Orchid is connected to your inbox and calendar, remembers you across conversations, and acts on your behalf. We put it next to the closest product that also lives in iMessage in [Orchid vs Poke](/compare/poke).

If you've read this far, the fastest way to understand it is to try it. Text [+1 (415) 605-3216](sms:+14156053216) and say hi 🌸

---

## Is Orchid AI Free? Pricing, Who Gets It Free, and What You Pay For

URL: https://orchid.ai/blog/is-orchid-ai-free-pricing-explained
Published: 2026-09-14
Topic: Guide

Short answer: no. Orchid is paid, from $29.99/mo, flat, with no free tier and no meter on your messages. Why we built it that way, and who does get it free.

"Is Orchid AI free?" is one of the most common things people search right before they text us. Here's the short answer, then the reasoning, then the exceptions.

No. Orchid is paid, from $29.99/mo, as a flat subscription. There's no free tier and no meter running on your messages. Students and YC founders get a free stretch, and we'll get to that.

## Why we didn't build a free tier

Almost every consumer AI product has a free plan. We chose not to, for two reasons worth saying out loud.

The first is that an assistant that reads your inbox shouldn't be paid for with your inbox. Orchid connects to your Gmail and your calendar. The only way to run that without ever selling data or training on it is to charge for it. Our [privacy promise](/legal/privacy) is credible because you're the customer, not the product.

The second is that the thing Orchid replaces isn't a chatbot subscription. It's the executive assistant you'd otherwise hire. We did that math in [The True Cost of Hiring an Executive Assistant in 2026](/blog/the-true-cost-of-hiring-an-executive-assistant-in-2026): salary, benefits, recruiting, ramp, turnover, and the three to five hours a week you spend managing them. Against that number, a flat monthly price is a rounding error, and that's the honest comparison.

## What the plan buys you

Everything Orchid does from your thread. Inbox triage and drafting, scheduling, meeting prep, reminders and follow-ups, voice notes, carousels, and the memory that ties them together. There's no per-feature upsell inside the assistant. If you want the full tour, it's in [Orchid on iMessage](/blog/orchid-ai-assistant-imessage).

The evidence that this kind of help pays for itself has gotten deep. A field study of about 5,000 support agents found that an AI assistant raised productivity 14% on average, with the biggest gains for the least experienced workers ([Brynjolfsson, Li and Raymond, NBER 2023](https://www.nber.org/papers/w31161)). A controlled experiment on professional writing found that access to ChatGPT cut time spent by roughly 40% while raising quality ([Noy and Zhang, Science 2023](https://www.science.org/doi/10.1126/science.adh2586)). And a study of 758 consultants at BCG found the group with AI finished 12% more tasks, 25% faster, at more than 40% higher quality on work inside the model's competence ([Dell'Acqua et al., Harvard Business School 2023](https://www.hbs.edu/faculty/Pages/item.aspx?num=64700)). Those were chat tools. An assistant that's wired into your actual inbox and calendar, and that acts instead of suggesting, compounds on top of that.

## Who does get it free

Students get Orchid free for 3 months, then 50% off a yearly plan. The details are at [orchid.ai/students](/students), and one student wrote up [what she did with it](/blog/how-i-used-ai-to-get-a-4-0).

YC founders get 3 months free. Email [nizzy@orchid.ai](mailto:nizzy@orchid.ai) with your name, company, and batch.

If you're leaving Notion Mail before it shuts down, we're applying a free month. Start at [orchid.ai/migrate/notion](/migrate/notion).

## Cancelling

From settings at [app.orchid.ai](https://app.orchid.ai), any time. Disconnecting your accounts cuts Orchid's access immediately, and you can delete everything from there too. No retention call, no "are you sure" email chain.

## Frequently asked questions

**Is there a free trial?**

Not a general one. Students and YC founders get 3 months free, and Notion Mail switchers get a free month. Everyone else starts on the paid plan and can cancel whenever.

**How much does Orchid AI cost?**

From $29.99/mo, billed monthly. Current plan details live inside the app.

**Does Orchid charge per message or per email?**

No. The plan is flat. A busy inbox doesn't cost you more.

**Is Orchid cheaper than an executive assistant?**

By a lot. A US executive assistant runs well over $70,000 a year once benefits and your own management time are counted; the [full breakdown is here](/blog/the-true-cost-of-hiring-an-executive-assistant-in-2026). Orchid isn't a great human EA for everything, but it covers the daily coordination work that eats most of their hours.

**Does Orchid use my data to keep the price down?**

No. We never sell your data, share it, or train on it. That's exactly why it's paid.

Want to find out if it's worth it for you? Text [+1 (415) 605-3216](sms:+14156053216) and hand it your inbox for a week 🌸

---

## Orchid AI Reviews, Reddit, and "Is It Safe?": Honest Answers

URL: https://orchid.ai/blog/orchid-ai-reviews-reddit-is-it-safe
Published: 2026-09-14
Topic: Guide

People search these before they text us. So here's what Orchid can see, what it can do without you, what happens to your data, and where the real reviews are.

Before people text an AI their inbox, they search. "Orchid AI review." "Orchid AI reddit." "Is Orchid AI safe." We see those queries land on our site every day, and until now the answers lived on other people's pages. So here's ours: what Orchid can see, what it can do without you, what happens to your data, and where to find real reviews.

## The right question to be asking

Being skeptical of AI assistants is the majority position, not the fringe. Pew found that 52% of Americans are more concerned than excited about AI's growing role in daily life, up from 37% two years earlier ([Pew Research Center, 2023](https://www.pewresearch.org/short-reads/2023/08/28/growing-public-concern-about-the-role-of-artificial-intelligence-in-daily-life/)). That concern is reasonable. Something that reads your email has access to more of your life than almost any other piece of software you use.

So we built Orchid around one rule. It should never be able to do anything you couldn't undo, and you should always be able to see exactly what it has access to.

## What Orchid can see

Only the accounts you connect. Today that's Gmail and Google Calendar, through Google sign-in. You never give Orchid a password.

Access to Gmail data is governed by Google's [API Services User Data Policy](https://developers.google.com/terms/api-services-user-data-policy), which restricts what an app can do with mailbox data and requires independent security review for apps that read email. Orchid operates under those terms. In plain English: we can use your mail to be your assistant, and for nothing else.

The conversation itself happens over iMessage, which is end-to-end encrypted between your phone and Apple's servers ([Apple Platform Security](https://support.apple.com/guide/security/imessage-security-overview-secd9764312f/web)).

## What it can do without you

By default, nothing that leaves the thread. Drafts stay drafts until you say send. Calendar holds wait for your okay. Orchid can read, summarize, prepare, and remind on its own, because none of those touch the outside world. Anything that does, waits.

You can widen that if you want, say, letting it confirm meetings with people you name, and you can pull it back at any time. The boundary is yours.

## What happens to your data

We never sell it, never share it, never train on it. You can disconnect any account or delete everything from settings, and revoking Orchid from your Google account cuts it off on Google's side the same instant. The full terms are in the [privacy policy](/legal/privacy) and the [terms of service](/legal/terms).

This is also why Orchid costs money. When the customer pays, the customer stays the customer. We explain the pricing in [Is Orchid AI free?](/blog/is-orchid-ai-free-pricing-explained)

## What people say after using it

The most useful reviews of Orchid are the ones people wrote themselves, about one specific thing it changed:

- A student who used it to [run a semester and finish with a 4.0](/blog/how-i-used-ai-to-get-a-4-0).
- Someone who [lost 15 pounds by texting an AI](/blog/i-lost-15-pounds-by-texting-an-ai) instead of opening another tracking app.
- A founder on [staying in touch with the people you mean to call](/blog/the-people-you-mean-to-call).

For a head-to-head against the closest product that also lives in iMessage, read [Orchid vs Poke](/compare/poke). It says where Poke is the better pick, because a comparison that never concedes anything isn't worth your time.

If you're looking for Reddit threads or independent reviews, go read them. We don't control them and we'd rather you show up informed. If you find something wrong or out of date, email [nizzy@orchid.ai](mailto:nizzy@orchid.ai) and we'll fix it or explain it.

## What AI assistants get wrong

Any honest safety page has to say this: language models make mistakes. They misread a date, summarize a thread with too much confidence, draft a reply that misses the tone. The research on people working with AI finds the gains are largest when the human keeps the judgment calls and the AI takes the drafting and the legwork ([Dell'Acqua et al., 2023](https://www.hbs.edu/faculty/Pages/item.aspx?num=64700)). That's the whole idea behind "drafts stay drafts." Orchid is built to make you faster at deciding, not to decide for you.

## Frequently asked questions

**Is Orchid AI safe to connect to my email?**

Orchid uses Google sign-in, never stores your password, operates under Google's API Services User Data Policy, and can't send mail or book meetings without your approval. You can revoke access any time from Orchid settings or straight from your Google account.

**Is Orchid AI legit?**

Yes. The founders answer at [nizzy@orchid.ai](mailto:nizzy@orchid.ai), and the product has been public since [March 2026](/blog/orchid-beta-is-here).

**Does Orchid read all my email?**

It reads the mail in the accounts you connect, so it can triage and draft for you. It doesn't read accounts you haven't connected, and it never uses any of it for anything other than helping you.

**Does Orchid train AI models on my data?**

No. Never.

**Can I delete my data?**

Yes. Disconnect accounts or delete everything from settings at [app.orchid.ai](https://app.orchid.ai). Revoking access from your Google account works immediately too.

**Where can I read independent Orchid AI reviews?**

Search for them; we don't curate them. Then text [+1 (415) 605-3216](sms:+14156053216) and write your own 🌸

---

## Introducing Keiki

URL: https://orchid.ai/blog/introducing-keiki
Published: 2026-08-25
Topic: Product

Build a capable customer-facing AI agent once, then make it available across iMessage, SMS, WhatsApp, Slack, Telegram, and email.

Great products reduce the distance between an intention and a result. Messaging is already where people ask questions, make decisions, schedule appointments, buy things, and get help. The natural place for a customer-facing agent is not another app or a widget waiting on a website. It is the conversation the customer already has open.

Keiki is a platform for building agents that live in those conversations. You define what the agent knows, how it behaves, what it remembers, which tools it can use, and where its boundaries are. Keiki provides the systems required to make that agent available across messaging platforms and operate it reliably.

The result is one agent that can represent your business wherever customers choose to reach it.

## Why we built Keiki

When we started Orchid, we wanted to create an AI assistant that was as easy to use as texting another person. It had to live in iMessage, understand the person it was speaking with, remember what mattered, and be capable of getting real work done.

The idea was simple. Building it took us almost a year.

The model could produce a useful response early on. Turning that response into a dependable product consumed everything that followed. We had to build messaging infrastructure, conversation management, durable memory, browser use, tool execution, background jobs, observability, failure recovery, and the systems required to scale all of it.

We were building two products at the same time: the Orchid assistant and the platform required to make Orchid possible.

That experience taught us something important. The agent should be unique. Its personality, knowledge, judgment, and capabilities are what make it valuable. The infrastructure surrounding it is necessary, but it should not have to be rebuilt every time someone has an idea for an agent.

Keiki is the platform we wish we had when we started Orchid. It packages the systems we spent the last year building so other teams can begin with the part that is actually theirs: the agent.

## One agent across every channel

Messaging infrastructure has traditionally forced teams to organize around channels. They build a Slack bot, add a separate SMS flow, and eventually create something else for WhatsApp. Each version develops its own instructions, integrations, memory, and behavior.

Keiki treats the agent as the source of truth and the channels as distribution.

You define the agent once. Adding a channel does not require rebuilding its identity or operating logic. The same knowledge, tools, behavior, and safeguards remain in place whether the conversation begins through iMessage, WhatsApp, Slack, Telegram, or email.

A channel should be a distribution decision, not a second product.

### Begin with what the agent should accomplish

Creating an agent in Keiki starts with a plain-language description of its purpose. You might ask it to answer product questions, check inventory, qualify potential customers, or schedule appointments.

The builder then helps you define what the agent needs. You can attach knowledge, connect services such as Notion, enable memory and browser use, add tools, and decide which messaging platforms it should support.

Before the agent speaks with a customer, you can review its configuration and test it in a private conversation. The goal is not merely to produce a quick demo. It is to shorten the path to an agent you can confidently put in front of people.

{{VIDEO}}

*Describe the agent, connect its knowledge and capabilities, select its channels, and test it before launch.*

### A useful agent has to do more than talk

A customer asking about an order does not need instructions for finding it. They want to know where it is. Someone trying to book an appointment does not want another link. They want the appointment booked.

Keiki agents can use a browser, search connected knowledge, call business tools, remember context, and perform work in the background. This allows them to move a conversation toward a result instead of stopping after an answer.

The example below shows a simple version of this. The agent is asked when Thanksgiving break begins, but the answer is not available in its knowledge. It opens the university calendar, finds the dates, and returns with the answer. The entire process is recorded and can be inspected afterward.

{{VIDEO}}

*When the answer was not available in its knowledge, the agent found it using a browser and returned with the result.*

Giving an agent the ability to act also requires control. Keiki lets teams decide which tools the agent may use, which actions require approval, and when a conversation should be handed to a person.

The quality of an agent is not measured only by how well it speaks. It is measured by whether it can reach the right result, remain within its boundaries, and recognize when human judgment is needed.

### Built to be operated

Launching an agent is the beginning of the product, not the end of the project.

A customer-facing agent needs to be observed, evaluated, and improved. Teams need to understand which conversations it resolved, where it struggled, why it made a decision, and when a person had to intervene.

Keiki brings conversations from every channel into one inbox. People keeps a record of everyone the agent interacts with. Knowledge contains the information available to it. Loops shows recurring work, while Metrics tracks usage, messages, and errors.

Every agent run also includes a readable record of what happened. You can inspect what the agent remembered, which information it retrieved, which tools it used, what it decided, and what each step cost.

This creates the operating loop required to build a genuinely good agent: observe what it does, understand why it did it, and improve it over time.

{{VIDEO}}

*A complete agent run, including every step, action, decision, and cost.*

## What you can build

Once an agent can combine conversation, memory, vision, a browser, and access to business systems, it can provide a complete service rather than perform a single task. With the appropriate tools, data, permissions, and human safeguards in place, teams could build agents like these.

**A travel company that lives inside the group chat.** A family invites the agent into the iMessage thread where they are planning a trip. It understands everyone's preferences, finds flights and hotels that satisfy the group, and builds an itinerary around the conversation. Once the travelers approve it, the agent makes the bookings. During the trip, it monitors changes, finds alternatives when a flight is cancelled, moves the hotel reservation, updates the calendar, and keeps everyone informed in the same thread.

**A personal shopper that turns a photograph into an order.** A customer sends a picture of a jacket they saw on the street. The agent identifies the style, searches the retailer's catalog, checks availability in the customer's size, and suggests alternatives based on previous purchases. It answers questions about fit and materials, assembles the cart, completes checkout after approval, tracks delivery, and handles an exchange if the item is not right.

**A mortgage advisor that takes a buyer from curiosity to preapproval.** A customer begins with a simple message about the kind of home they hope to buy. The agent explains the tradeoffs, gathers the necessary information over time, retrieves credit with permission, presents personalized rates, and helps the customer understand the available loan options. When the moment calls for a licensed professional, it brings in the right person with the entire conversation and application already organized.

**An insurance and repair concierge that makes a damaged windshield disappear as a problem.** A driver sends a photograph of a crack and their license plate. The agent identifies the vehicle, explains the relevant insurance coverage, begins the claim, determines which glass and camera calibration are required, and finds an appointment that fits the driver's schedule. It arranges a mobile repair, sends updates, and remains available until the car is ready.

**A veterinary front door that is available the moment something goes wrong.** A pet owner messages that their dog swallowed something it should not have. The agent gathers the animal's age, weight, symptoms, and timing, then provides guidance grounded in the clinic's approved medical protocols. It identifies whether the situation requires immediate attention, books the appropriate appointment, and gives the veterinary team a structured summary before the owner arrives. Routine questions are resolved quickly; uncertain or urgent cases go directly to a professional.

None of these experiences is simply a chatbot. Each is a service delivered through conversation.

When an agent can remember, reason, use software, and continue working after a message has been sent, the messaging thread stops being a support channel. It can become the product.

{{VIDEO}}

*A Keiki agent answering for a ramen shop in iMessage: the story, the hours, and what to order.*

## How it works

Most teams can build their agent through the visual builder. They describe its purpose, connect knowledge and tools, select the channels, test the result, and launch it. Keiki manages the infrastructure behind it.

Developers can define the same agent in code, add custom tools, and control how it is deployed.

```ts
import { create } from '@orchid/agents'

const kit = create({
  apiKey: process.env.ORCHID_API_KEY!,
  name: 'Support Agent',
  agent: {
    model: 'google/gemini-3.5-flash',
    systemPrompt: ({ now }) =>
      `You are the support agent for Acme. Current time: ${now.toISOString()}`,
    tools: [lookupOrder, bookAppointment],
  },
})

await kit.sync()
```

`kit.sync()` gives the agent to Keiki to run.

Teams that need complete control or have strict data requirements can run the agent on their own infrastructure. The way they build it remains the same.

### Works with the AI tools you already use

Keiki agents also speak the APIs that AI software already understands.

Any application that connects to OpenAI or Anthropic can connect to your agent. Point it at your Keiki address, use the agent's name as the model, and each completion becomes a conversation with that agent.

```ts
import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://<host>/api/v1/openai',
  apiKey: process.env.ORCHID_API_KEY,
})

const reply = await client.chat.completions.create({
  model: 'support',
  messages: [{ role: 'user', content: 'where is my order?' }],
})
```

It is still the same agent, with the same knowledge, memory, tools, and rules. The API is another way to reach it.

## Who Keiki is for

Keiki is for founders and operators who have an idea for an agent but do not want to spend a year building the infrastructure required to make it real.

It is for product and customer-experience teams that want AI to handle customer conversations with useful context, controlled actions, clear escalation, and full visibility.

It is also for developers who want the speed of a platform without giving up the ability to write their own tools, define the agent in code, or control where it runs.

A static FAQ widget does not need Keiki. Keiki is for teams building agents that are expected to represent the business, use judgment, and complete meaningful work.

## The larger idea

The dominant interface of each computing era has moved closer to the way people naturally think.

Command lines required people to learn the language of computers. Graphical interfaces gave them familiar objects. Touch made software direct and physical. Conversation allows people to begin with their intention.

That shift changes what software can be. A business will no longer be represented only by a place people visit. It can also be represented by an agent people speak with: one that understands the business, remembers the relationship, and can help accomplish something.

Software is becoming someone.

Keiki is how you decide who that someone is, what it knows, what it can do, and where people can reach it.

## Build your first agent

We spent almost a year building the foundation that made Orchid possible. Keiki gives other teams a way to begin with that foundation already in place.

Start with the agent you wish existed. Define what makes it valuable, connect what it needs, and make it available where the conversation is already happening.

**[Build your first agent →](https://onkeiki.com/signup)**

## Frequently asked questions

**What is customer-facing AI?**

Customer-facing AI refers to AI agents that customers interact with directly: answering questions, taking actions, and completing transactions in the channels customers already use, such as SMS, WhatsApp, and email. Keiki is a platform for building and operating these agents.

**What is Keiki?**

Keiki is a platform for creating AI agents with a role, memory, knowledge, and tools, and deploying them across messaging channels. It handles the runtime, transport, billing, observability, and execution so teams can focus on what the agent knows and does.

**What channels does Keiki support?**

SMS, iMessage/RCS, WhatsApp, Slack, Telegram, and email, plus OpenAI- and Anthropic-compatible API endpoints. Channel adapters are extensible rather than a closed list.

**Can Keiki connect to my existing systems?**

Yes: through custom tools, MCP servers, installable capabilities, HTTP integrations, browser automation, and sandboxed code execution.

**How do I control what the agent does?**

You define the agent's behavior and boundaries, and nothing runs until you review and create it. New tools ask for approval by default, every conversation and run is inspectable down to the individual step, and sensitive actions can require a human's sign-off.

**Is Keiki just a chatbot?**

No. Keiki agents remember context across conversations, call tools, execute workflows, and complete transactions, and escalate to a human when they should.

---

## How I Used AI to Get a 4.0

URL: https://orchid.ai/blog/how-i-used-ai-to-get-a-4-0
Published: 2026-08-21
Topic: Story

Every August I built the perfect semester system, and every year it was dead by week three. Last spring I texted my syllabi to an AI instead. I never missed a deadline, and I ended the semester with a 4.0. The setup, in time for this fall.

Every August, I build The System.

You know The System. The Notion dashboard with a page per class. The color-coded calendar. The planner app with the streak counter. One year it was a bullet journal with hand-drawn habit trackers. The System is beautiful, and it works for about nineteen days.

Then a professor moves the midterm. A problem set drops on a Tuesday instead of a Thursday. A club trip eats a weekend. The System quietly stops being true, and a system that is 90% true is worse than no system at all, because now checking it feels like lying to yourself. By week five I'm running the semester on vibes, screenshots, and fear.

## Why every system died

It took me three years to see it: I was The System's only employee.

I typed in every syllabus date. I caught every "check the portal" tossed out at the end of lecture, remembered it, moved it. The dashboard didn't track my semester. It tracked my willingness to do data entry at 11pm.

## So this time I photographed the syllabi

Last spring, during syllabus week, I tried something different. Five classes, four PDFs and one crumpled paper handout. I photographed all of them and texted them to Orchid, the way you'd forward them to a scarily organized friend.

A minute later:

*got all 5. i count 61 dated things: problem sets, quizzes, two essays, labs, midterms, finals. want everything on your calendar, or just reminders for the ones that can hurt you?*

Then it asked the two questions a person would ask. My econ syllabus said "weekly quiz, day announced in lecture," so instead of guessing, it offered to check with me every Monday. My lab section wasn't assigned yet, so it left those slots open and told me to text it when registration cleared.

That was the entire setup, no template, no dashboard. The thread was the system now.

## The week-three test

In February, my stats professor ended a lecture with "we're pushing the midterm a week, check the portal." That sentence killed every system I ever built, because it creates work at the exact moment I can't do it: packing up, half out the door, already late.

This time I texted one line from the lecture hall: *stats midterm moved a week*.

*updated to march 6. i shifted both review blocks and the night-before reminder with it. old date is off your calendar.*

The moment the semester used to start rotting is now a text I send from my seat.

## It texts first

Every app I quit was patient. It waited for me to open it, which, in the exact weeks I was drowning, is what I stopped doing. Orchid doesn't wait.

The night before a deadline: *gov essay due tomorrow at 11:59pm. you haven't mentioned it all week. still on track?*

Sunday evening: the week ahead in one text. What's due, which days are heavy, where the free window is.

And every morning, a short brief: today's classes, the deadline that matters, the one email that actually needs me.

I handed it my gym schedule too, three mornings a week ([someone around here lost 15 pounds this way](/blog/i-lost-15-pounds-by-texting-an-ai)). When I skip, it notices: *nothing logged yet this week, tomorrow at 8?*

The worst part of any semester was the low hum of something, somewhere, being due, and me not knowing what. The hum is gone.

## The .edu inbox is a firehose

My school inbox gets forty emails a day, and three a week matter. Club listservs, career fair blasts, dining hall surveys, and buried in the pile: the TA moving office hours, the recruiter replying, the registrar deadline with real consequences.

Orchid reads all of it and surfaces what needs me. *your TA moved thursday office hours to 4pm, want it on your calendar?* The recruiter's reply arrived as a text with a draft response attached. I read it, said send, kept walking to class.

I also help run a club, which is mostly chasing people. Room bookings, the sponsor who never answers, the treasurer who needs reminding twice. Orchid drafts the chase, I approve it, and I stopped being the group chat's most annoying presence.

## What it didn't do

It never wrote an essay. It didn't take a quiz for me. I wouldn't want it to; outsourcing the learning is failing out with better logistics.

What it did was quiz me from my own notes the night before (*ok, chapter 4. what breaks when you assume the samples are independent?*), turn three weeks of lecture notes into a summary I could reread over breakfast, and remember in April what the professor had said in February about the final's format, when everyone in the room, including me, had forgotten.

The learning stayed my job. Everything around the learning stopped being my job. That's the trade I spent three years trying to make with dashboards.

My semester GPA was a 4.0, my first ever. I want to be precise about what the AI earned there: none of the points and all of the room. I didn't miss a single deadline, I walked into every class knowing what was due, and finals week, for the first time since freshman year, was just studying.

## The fall setup takes two minutes

I did it again last week for this semester. Photograph the syllabi, send them, answer two questions. Done before the add/drop deadline, which it had already put on my calendar.

If you're a student, this is the cheapest a real assistant will ever be: Orchid is free for 3 months, then 50% off yearly with the student deal. Email [nizzy@orchid.ai](mailto:nizzy@orchid.ai) with your name and school, or start at [orchid.ai/students](/students).

Bring your syllabi 🌸

---

## Some answers don't fit in a bubble

URL: https://orchid.ai/blog/introducing-carousels
Published: 2026-08-06
Topic: Product

Carousels are live on Orchid. When an answer has real parts, a trip itinerary, a headphone comparison, your workout split, Orchid sends a stack of photo cards you swipe through instead of a wall of text.

Some answers don't fit in a text bubble. A weekend itinerary, a headphone comparison, a workout plan: the honest answer has five or six parts, and in a chat thread that becomes a wall of gray text you scroll past twice and never quite absorb.

So we taught Orchid to send cards instead. When an answer needs a little more than just some text, Orchid designs a stack of four to six full-screen cards and sends it as one message. Each card gets a photo Orchid went and found, a short header, and a couple of sentences that lead with the point. You swipe through it like a story.

We built the first version for meetings: a briefing before your next call with who you're meeting, what happened last time, and what to say. People loved swiping through a briefing much more than reading one, and we kept wanting the same treatment for everything else. So we generalized it. Orchid now decides on its own when an answer deserves cards, and it can build them for anything.

## Plan a trip you can hold in one hand

_hey orchid! plan a trip to lisbon for 4 people this weekend, budget doesn't matter!_

What comes back is a stack: an opening card that sets the plan, fly in Friday, stay in Príncipe Real or Chiado, then a card per day, with real advice baked in, like booking dinner and any Fado you care about now because August runs hot and crowded. Every card sits on a photo of the actual place, so by the time you've swiped through, you've half-seen the trip.

And the cards are a starting point, not a brochure. Under the stack, Orchid keeps going: _send me your departure city and i'll lock the flight, hotel, transfers, and reservations for all 4 of you._

{{VIDEO}}

## Compare before you buy

Same shape, different job. Ask _can you compare the new airpods max and the sony xm6?_ and each contender gets its own card: the specs that matter, the weight difference in grams, the battery in hours. The last card is Orchid's pick, with the trade-off stated plainly.

You could assemble the same answer from an hour of reviews and open tabs. The carousel is much easier to parse, and it stays in your thread for when you're ready to order.

{{VIDEO}}

## A workout split you'll actually open

Tell Orchid you can train three days a week and you're chasing muscle, and the plan arrives as cards: one card per training day with the exact exercises, sets, and reps, and a final card on how to grow, down to the protein target per kilogram of bodyweight.

A PDF training plan dies in your downloads folder. This one lives in the same thread where [Orchid already logs your workouts](/blog/i-lost-15-pounds-by-texting-an-ai), two scrolls away when you're standing in the gym.

{{VIDEO}}

## How to get it

There's nothing to set up. If you text Orchid, you already have carousels: ask a question with parts, and Orchid reaches for cards when they beat plain text. You can also ask outright, _send it as cards_, and it will.

One note on where you're texting from: on iMessage the cards arrive as a single swipeable stack; on regular SMS and RCS they arrive as individual images, same content, less magic.

If you're new here, Orchid is an assistant that lives in your texts. No app, no dashboard. It handles your inbox, your calendar, your plans, and now it shows its work in a form you'll actually look at. [Say hi](https://app.orchid.ai?source=site&utm_source=site&utm_campaign=blog_carousels) and ask it to plan your weekend 🌸

---

## I Lost 15 Pounds by Texting an AI

URL: https://orchid.ai/blog/i-lost-15-pounds-by-texting-an-ai
Published: 2026-07-21
Topic: Story

I've quit more meal-tracking apps than I can count. The one thing that stuck wasn't an app at all. It was a text thread that remembered my goal, logged my meals from a photo, and checked in on me when I needed it.

I've downloaded MyFitnessPal three times. I've quit it three times.

Not because it doesn't work. The problem was me. Logging a meal meant unlocking my phone, finding the app, waiting for it to load, searching a database for "chicken burrito bowl," and picking the right one out of forty. By the time I'd done that twice I'd decided I'd catch up later, and later never came. By day four the app was a guilt icon I swiped past.

The thing I actually needed was never more information about my food. It was consistency. Something so frictionless I couldn't talk myself out of it, that noticed when I slipped.

So a few months ago I stopped opening an app. I just started texting.

## The setup took about thirty seconds

I texted Orchid the way I'd text a friend: _i want to lose some weight, can you help me track my meals and keep me accountable?_ No sign-up flow, no wizard, no "create your profile." It asked what I was aiming for and how often I wanted to weigh in, then set it up: a running food log and a Sunday-morning weigh-in. _send me your meals whenever, a photo is easiest._ That was the whole onboarding. It lives in my texts now, next to my actual friends.

{{VIDEO}}

## The habit that finally stuck: snap and forget

Here's what made it work. iMessage is already open. So logging a meal became: photo the plate, send it, put the phone down. _logged: salmon, rice, side of greens. clean lunch._

{{VIDEO}}

Sometimes it had an opinion, which is the part I actually needed. I'd send a photo of a giant burrito and chips and get back: _logged. that's a big one, go lighter at dinner and you're totally fine for the day._

I want to be honest about what this is. Orchid is not a nutrition database. It isn't spitting out "624 calories, 38g protein" with decimal-point confidence, and I didn't want it to. That precision was always a little fake, and chasing it was half of why I quit the apps. What it does is keep an honest running picture of how I'm eating and nudge me when a day is getting away from me. That was enough.

## The gym got the same treatment

Losing weight was never only about food. The other half was showing up to the gym, and the gym has the exact same problem: the hard part isn't knowing what to do, it's going on the days you'd talk yourself out of it.

So I gave Orchid my workouts too. After a session I'd text it what I did, like _push day, bench and shoulders, 45 min_, and it kept the log. When I asked, it could tell me how the week was going: _that's three sessions this week, one more and you hit your goal._

Then I asked it to keep me honest. Now on a day I haven't moved, it nudges me: _no workout logged yet. even twenty minutes counts._ It's not a training plan and it's not a wearable. It's the same thing that worked for meals: something that remembers what I'm aiming for and won't let me quietly skip.

{{VIDEO}}

## The moment it clicked: it texted me first

For the first week it was a better logging app. Then, on a Friday I wasn't thinking about any of this, my phone buzzed: _hey, you logged 5 days straight this week and you're down 2 lbs. whatever you're doing, keep going._

{{VIDEO}}

I didn't ask for that. I didn't open anything. It came to me.

That's the part no app has ever done for me, because an app can't. An app waits for you to feel motivated enough to open it, which, when you're struggling, is exactly when you don't. Orchid flipped it. It remembered my goal, noticed I was showing up, and reached out at the moment a little encouragement actually mattered. It stopped being a tool I used and became something closer to a coach I had.

## Fifteen pounds later

I'm down about 15 pounds.

The lesson isn't that AI has a magic calorie model. It doesn't. What changed is that the friction disappeared and the accountability didn't. Logging got so easy I stopped skipping it, and on the days my motivation dipped, the one thing that had always been missing showed up on its own and told me I was doing fine.

Every diet tool I'd tried made me do the remembering, the opening, the showing up. This one did the remembering and the showing up for me. That turned out to be the whole game.

## Try it

Orchid is an assistant that lives in your texts. No app to download, no dashboard to check. It remembers what matters to you and actually reaches out. This is just the thing I used it for; people use it for their inbox, their calendar, their whole day.

If you've quit more tracking apps than you'd like to admit, this is the one that's just a text away. 🌸

---

## The People You Mean to Call

URL: https://orchid.ai/blog/the-people-you-mean-to-call
Published: 2026-06-30
Topic: Product

Orchid remembers the people who matter to you. Their birthdays, what they care about, when you last spoke. So you can stay in touch without keeping it all in your head.

There's a particular kind of guilt that lives quietly in everyone. The friend you keep meaning to text back. The mentor who changed your life, who you haven't thanked in years. The birthday that slipped past while you were heads-down at work, so you sent the slightly-too-late "happy belated!" again. You don't love these people any less. You just lost the thread, and the longer it stays lost, the heavier it gets to pick back up.

We know that feeling well, because we lived it. We're builders, we're busy, and somewhere in the noise we became people who let good relationships go quiet. Not out of coldness, just out of forgetting. So we built Orchid to fix it.

## Not a contact list. Someone who remembers with you.

Most tools made for this turn people into rows. Add a contact, set a reminder, log the call, search the spreadsheet for whose birthday is coming up. It's a database wearing a friendly face, and it makes you feel less like a friend and more like an account manager working a pipeline.

Orchid works nothing like that. You don't maintain a personal CRM. It pays attention to what's already happening in your life, the dinners on your calendar, the dates and details people mention, and holds onto what you'd want to hold onto. That his birthday is Thursday. That her dad's surgery is this week. That you promised to send that book and never did.

Then, at the right moment, it taps you on the shoulder. _It's Marcus's birthday tomorrow, and his big presentation was today. Want to check in?_ You say yes. The message goes. You never had to carry it.

{{VIDEO}}

## Never miss a birthday, or anything else that matters

Orchid makes sure you never miss a birthday again. No last-minute scramble, no "happy belated," no leaning on a social network to tell you who to care about today. You get a warm nudge in advance, with time to actually do something thoughtful.

And birthdays are just the start. Orchid also remembers the follow-ups you meant to send, the personal details that make a message land, and the quiet sense of when it's been too long with someone. Tell it once and it never forgets. So when you reach back out, you're not starting cold. You're walking right back in like you never left.

{{VIDEO}}

## Who this is for

If you've ever wanted to be the person who always remembers birthdays. If you've moved cities and watched your friendships thin out. If you have a parent you don't call enough or an old friend you keep meaning to reconnect with. If you've ever wished you were better at staying in touch, not because you don't care but because life got loud, this is for you.

It's early, but it works, we use it every day, and it's made us more present in the lives of the people we love. Connect your accounts and Orchid starts learning in minutes.

Some things are too important to leave to memory. 🌸

---

## The Next App Is No App

URL: https://orchid.ai/blog/the-next-app-is-no-app
Published: 2026-06-18
Topic: Essay

For thirty years the deal was simple: capability in exchange for your attention, one screen at a time. The next app isn't a smarter app or a better grid. It's a message, in the thread you never close.

If you opened your phone right now and counted the apps, you might find forty. Maybe eighty. Maybe two hundred.

Every one of them was a promise. Download me and your life gets easier. And every one of them came with the same fine print: first, learn everything about me.

We accepted this. We accepted that wanting to do something meant first finding the right rectangle on a grid, opening it, remembering how it works, and navigating to the part that does the thing. A different app for the flight. A different app for the bank. A different app for the food, the ride, the calendar, the doctor, the dry cleaner.

We naively called this convenience, but it was just a filing system for intentions.

## The app was always a workaround

We honestly learned this the hard way.

Some of you might remember 0.email. It was an email client, and it was a really good one. But it became really clear that it is just a middleman between you and your goals. It became this monster we 1) couldn't tame, and 2) couldn't make an impact on the world with. Then we built orchid.ai, a different take on what an email client is. Also good, but in the end, just a middle man between you and your objectives. It was one more thing asking you to come to it, learn it, operate it. No matter how perfectly we nailed the UI, or how well thought out our experiences were, we had this feeling that something was still missing.

Looking back, it's clear that that feeling was that we never wanted another app.

Each time, we'd built another workaround and called it a product. The thing we kept making was the very thing people didn't want: a place to go, a tool to wield, a manual to memorize. It took us a while to see that we were solving the wrong problem. The problem was never that the apps were bad. It was that there was an app at all.

Well… what did we want? We wanted to book the flight. We wanted to handle the refunds. We wanted dinner at 8...

## So what replaces it

Not a better app. Not a smarter grid. Not an app that finally, after all these years, has good search.

A conversation.

We already know how to do this. We've been doing it our whole lives. You tell a competent person what you want, and they handle it. You don't open them. You don't navigate them. You don't learn their interface. You just say the thing.

"Book me Aruba, soon."

"What was that restaurant we loved last March?"

"Remind me to take my meds at 9."

There's no screen to learn here because there's no screen. There's intent, and there's outcome, and nothing in between that you have to operate. The interface disappears, because the best interface was never an interface. It was someone who already understood you.

## Why messages?

If the next app is no app, it has to live somewhere. And the somewhere is already on your phone, already open, already the most-used thing you touch all day.

Your messages.

Think about what the messaging thread actually is. It's the one place on your phone with no learning curve, because it's the purest form of human expression. It's where your real life already happens, the group chats, the plans, the people who matter. It's the app you check first in the morning and last at night, not because anyone gamified it, but because that's where the people are.

Now imagine an assistant there. Not as another icon to find. As another thread. The same place you text your partner, you text the thing that books the flight, tracks the calories, drafts the reply, catches the deadline before it slips. No new app to download. No new place to check. No new behavior to build.

And it's everywhere you are. Your phone, your watch, your laptop, the same thread following you across all of it. No context to rebuild. No tab to find. The assistant lives in the channel you never close.

The app asked you to come to it. The thread is already where you are.

## The quiet part

This isn't a smaller version of the app era. It might be the end of it.

For thirty years the deal was: we give you capability, you give us your attention, one screen at a time. The app store was the field of that deal; discovery, downloads, icons, badges, with the express purpose of capturing your attention.

The next era doesn't have icons. It has a thread, and on the other end of it, something that does the work the app used to make you do.

You probably won't miss the apps. The same way few people miss the filing cabinet, or the phone book, or the travel agent's printout. You'll just notice, one day, that you've stopped opening things. That you started asking instead. That the gap between wanting and having got so small you can text it.

The next app is no app.

It's a message.

---

If your phone has started to feel like a grid of chores you have to operate yourself, we built Orchid for exactly this. One thread, in the messages you already use, that handles the rest. [Say hi](sms:+14156053216) and let it run 🌸

---

## The World Cup, In Your Messages

URL: https://orchid.ai/blog/the-world-cup-in-your-messages
Published: 2026-06-11
Topic: Product

Our World Cup experience is live. Text Orchid and get live scores, goals, red cards, and final whistles for every match. And if you're going to a game, it doubles as the best local guide in town.

The World Cup starts today.

48 teams. 104 matches. 16 host cities across the United States, Canada, and Mexico. The biggest tournament ever played, stretched across five weeks and four time zones, and most of it will happen while you're in a meeting, on a flight, or at a wedding you couldn't get out of.

So we built something for it. Starting today, Orchid follows the World Cup with you. Text [+1 (415) 605-3216](sms:+14156053216) and you're in. No app to download, no account to set up, no notification settings to tune. It's a text thread, like the ones you already have.

## Every goal, the second it happens

Tell Orchid which teams you care about, or tell it to keep you posted on everything. From kickoff, the match comes to you:

> ⚽ GOAL. Messi, 23'. Argentina 1-0

Goals. Red cards. Halftime scores. Final whistles. The moment they happen, not twenty minutes later when you finally check your phone.

And because it's a conversation, not a notification, you can text back. Ask how a player is doing and you'll get a real answer ("unstoppable. 4 shots, 1 goal, 38 minutes in"). Ask what you missed while you were heads-down. Ask if the penalty was soft. Orchid is watching the match either way.

## Never miss a kickoff

A tournament played across four time zones is a scheduling problem, and scheduling problems are what Orchid is for.

Tell it "wake me up for the final" and it will text you 30 minutes before kickoff, in your time zone, with the lineups as soon as they drop. Miss a match entirely and you get the recap: the final score, the moments that mattered, and what it means for who advances.

## If you're going to a match

This is the part we're most excited about.

Millions of fans are traveling to host cities this summer, and most of them are landing somewhere they've never been. This is where Orchid stops looking like a scores feed and starts being what it actually is: an assistant.

The thing on the other end of the thread isn't a sports bot. It's the same personal assistant that books trips, keeps calendars in order, and never forgets a detail. It can search, plan, and coordinate. So when you land in Mexico City the morning of a match, you can text it the way you'd text a friend who lives there:

"where should we eat near the stadium tonight, party of 4"

"we've got a free day before the quarterfinal, what's actually worth doing"

"find a bar near our hotel showing the early match"

Restaurants worth the wait, neighborhoods worth the detour, the smart way to get to the stadium, what to do with the dead day between matches. Orchid figures it out and texts you a plan.

## Why we built this

Orchid is a personal assistant. It helps you stay organized, find information, and get things done, all through messages. Most of what it does happens quietly: booking the trip, sorting the admin, remembering the detail you mentioned once. Useful, but hard to show someone at a party.

The World Cup is the opposite. It's a month where everyone, everywhere, wants the same thing at the same time: to not miss the moment. And we think the fastest way to understand what having an assistant feels like is to delegate something you actually care about and watch it get handled.

So for the next five weeks, this is the demo: a free kick in the 89th minute, in your messages, the second it hits the net.

The tournament runs through the final on July 19. Text [+1 (415) 605-3216](sms:+14156053216) or visit [orchid.ai/worldcup](https://orchid.ai/worldcup) to get started.

We'll see you at the final.

---

## Orchid + Granola

URL: https://orchid.ai/blog/orchid-and-granola
Published: 2026-06-01
Topic: Product

Our Granola integration is live. Meeting notes, action items, and decisions now flow straight into Orchid, so the most valuable context in your company doesn't disappear the moment a meeting ends.

Today, we're launching our Granola integration.

Meeting notes, action items, decisions, and context can now flow directly into Orchid, helping it better understand what you're working on and what deserves your attention next.

On the surface, this looks like another integration. We think it's much more important than that.

Meetings contain some of the most valuable context in a company. New projects are discussed. Customers share feedback. Decisions get made. Priorities change. Relationships deepen. Entire strategies can shift in a single conversation. Yet most of that context disappears the moment the meeting ends.

Notes get saved. Recordings get archived. Action items get forgotten. Valuable information becomes trapped inside another application, disconnected from the rest of the work that follows.

We've spent a lot of time thinking about this problem at Orchid because a surprising amount of modern work isn't the meeting itself. It's everything that happens afterward.

## Better meeting prep

One of our favorite use cases is meeting preparation.

Most people walk into meetings with partial context. They vaguely remember who they're talking to, what was discussed last time, and maybe a few notes if they had time to prepare. The reality is that most founders, operators, salespeople, recruiters, and managers are juggling dozens or hundreds of conversations at once. Remembering every detail simply isn't realistic.

With Granola connected to Orchid, meeting context becomes part of the preparation process. Before a call, Orchid can pull together information from previous meetings, emails, calendar events, and conversations to create a briefing that helps you walk in prepared.

Who are you meeting? What are they building? What did you discuss last time? What commitments are still open? What should you bring up?

The best conversations happen when people feel understood. The challenge is that understanding people requires context, and context is often scattered everywhere. Instead of digging through notes, searching your inbox, or trying to remember what happened three weeks ago, Orchid can bring that information together automatically.

{{VIDEO}}

## Follow-ups are where work happens

Most meetings don't create value by themselves. The value comes from what happens afterward: the introduction you promised to make, the proposal you said you'd send, the customer concern you agreed to investigate, the candidate you wanted to follow up with, or the decision that needs to be communicated to the rest of the team.

These things sound simple, but they're often buried beneath the next meeting, the next notification, or the next urgent task. We've found that many of the most important opportunities aren't lost because people don't care. They're lost because people are busy.

When Orchid understands what happened during a meeting, it can help ensure the work continues moving forward afterward. Follow-ups become easier to identify, easier to prioritize, and harder to forget. Instead of notes sitting passively in a document, they become part of an active workflow that helps move work forward.

{{VIDEO}}

## Action items shouldn't disappear

Every meeting creates work, but action items frequently become trapped inside meeting notes.

Someone volunteers to own a task. A deadline gets discussed. A follow-up gets mentioned. Everyone agrees on next steps. Then the meeting ends, and a week later nobody remembers exactly who was responsible for what.

This is one of the most common coordination failures inside organizations. Not because people aren't capable, but because modern work generates more commitments than any individual can realistically keep track of.

By bringing meeting context into Orchid, action items can become part of a broader understanding of what needs to happen next. Important commitments don't have to live exclusively inside a note-taking application. They can become part of the systems people already rely on every day, making them easier to surface, prioritize, and execute on.

{{VIDEO}}

## Context compounds

One thing we've learned while building Orchid is that assistants become more useful as they gain context.

Your emails contain context. Your calendar contains context. Your meetings contain context. Each source tells part of the story, and when those pieces come together, the assistant develops a much better understanding of what you're working on, who you're working with, what commitments exist, and what deserves your attention right now.

This is why we get excited about integrations like Granola. It's not because another application was connected. It's because Orchid gains access to another layer of understanding.

The more complete the picture, the more helpful an assistant can become.

{{VIDEO}}

## Most work doesn't happen in a single app

A conversation becomes a meeting. The meeting creates action items. The action items generate emails. The emails lead to decisions. The decisions create new meetings.

Modern work flows across systems continuously, which means no single application contains the full picture.

We don't think assistants should be limited to a single application either.

The future isn't an assistant that lives inside your inbox, your notes, or your calendar. It's an assistant that understands how work moves between all of them.

The Granola integration is another step toward that future. Because meeting notes shouldn't disappear after the meeting ends.

We're building Orchid for people who want to spend less time coordinating work and more time doing it.

Join the beta at [orchid.ai](https://orchid.ai).

---

## Knowledge Work Is Becoming Coordination Work

URL: https://orchid.ai/blog/knowledge-work-is-becoming-coordination-work
Published: 2026-05-28
Topic: Essay

Models keep getting smarter, but people still feel overwhelmed. The bottleneck in modern work is no longer intelligence. It is coordination, the constant work of moving context between people, systems, and decisions.

We've spent the last few months thinking deeply about how humans and software will work together.

One thing became clear very quickly: the hard part is not intelligence.

Models are getting smarter every month. They can write, summarize, search, reason, code, and increasingly take action across software. But despite all of that progress, people still feel overwhelmed by work. In many ways, software has become more capable while humans have become more fragmented.

The reason is that most modern work is no longer constrained by access to information. It is constrained by coordination.

A huge percentage of knowledge work now consists of moving context between people, systems, and decisions. Following up on threads, preparing for meetings, remembering what was decided last week, updating the right tools, checking whether someone replied, finding the right document before a call, keeping projects moving across teams. None of these tasks are individually difficult, but together they create a constant background layer of mental overhead that consumes an enormous amount of attention.

## Software created the coordination problem

Every generation of software improved one specific part of work. Email improved communication. Calendars improved scheduling. Docs improved collaboration. Slack improved conversation. CRMs improved visibility. Project management tools improved coordination.

But every new tool also created another place where work could live, another surface that needed to be monitored, updated, and checked throughout the day.

Now important context is scattered across inboxes, chats, meetings, recordings, tickets, notes, dashboards, and dozens of disconnected systems. The problem is usually not whether the information exists. It is understanding where it lives, whether it matters, who needs to see it, and what should happen next.

Today, humans are still responsible for stitching all of that together manually.

Most people are not overwhelmed because they lack tools. They are overwhelmed because every tool competes equally for their attention.

## Coordination is becoming the real layer of work

Most software still treats work as isolated actions inside isolated apps.

Send the email.

Update the task.

Schedule the meeting.

Reply to the message.

Review the document.

But real work rarely happens that cleanly.

A customer conversation becomes a meeting. The meeting becomes a document. The document creates tasks. The tasks generate updates. The updates trigger new conversations. Important decisions get spread across inboxes, calls, notes, chats, and project boards.

Modern work is not linear anymore. It flows continuously across systems.

That is why coordination has quietly become one of the largest forms of work itself.

People spend enormous portions of their day trying to maintain continuity across fragmented systems. Checking whether someone responded. Remembering who owns what. Looking for context before meetings. Following up on stalled threads. Updating tools manually so other people stay informed.

None of this is the core work people actually want to be doing.

It is the overhead surrounding the work.

## Attention is becoming the bottleneck

The bottleneck in modern work is no longer capability. It is attention.

Every unread message feels urgent. Every notification asks for context switching. Every tool creates another stream of information competing for focus. The result is that software becomes increasingly capable while the human operating it becomes increasingly fragmented.

We think the next generation of software should not increase the amount of information users consume. It should reduce the amount of coordination work required to operate effectively in the first place.

That changes how systems should behave.

A useful system should understand urgency, timing, relationships, and context well enough to know when something deserves interruption and when it does not. Sometimes the right behavior is surfacing information immediately. Sometimes it is waiting until later. Sometimes it is taking action automatically. Sometimes it is intentionally staying silent.

The goal is not maximum visibility.

The goal is helping people focus on what actually matters while the coordination layer operates more quietly in the background.

## Building Orchid around coordination

This shift in how work happens has shaped a lot of how we think about Orchid.

Most systems today are optimized around capability. Can the model generate the response, complete the task, or answer the question?

Those things matter, but we think coordination quality matters just as much.

A useful assistant needs to understand more than the immediate request in front of it. It needs context around the broader flow of work. Who is involved, what matters to the user, what has already happened, what is blocked, what can wait, and what should never have interrupted the user in the first place.

That changes how the system behaves.

In Orchid, we spend a large amount of time thinking about prioritization, timing, memory, context, and attention management. The goal is not simply generating responses. It is helping users maintain continuity across the constant movement of modern work.

That means the system has to make judgment calls continuously.

Sometimes the correct behavior is drafting a follow-up before the user remembers. Sometimes it is preparing context before a meeting. Sometimes it is identifying that a conversation has stalled. Sometimes it is updating the right systems automatically. Sometimes it is deciding that something simply does not deserve the user's attention at all.

As systems become more autonomous, we think this behavioral layer becomes increasingly important. The quality of the experience is no longer defined only by whether the system can perform a task. It is defined by whether the system consistently demonstrates good judgment while operating across the user's workflow.

A large part of building assistants is not teaching systems how to generate better outputs. It is teaching them how to behave responsibly around human attention.

## The interface for work is changing

Today, most software products are destinations that users constantly bounce between throughout the day. You open the inbox, check the calendar, review the project board, search for the document, respond in Slack, update the CRM, then repeat the cycle again an hour later.

But the actual work rarely belongs to one tool.

We think the future interface is not another standalone app people need to constantly monitor. It is a coordination layer that understands the flow of work across the systems people already use.

That might mean preparing someone before a meeting, drafting follow-ups afterward, identifying stalled conversations, reminding the right person at the right time, updating systems automatically, or surfacing information before it is explicitly requested.

Not as separate automations stitched together manually, but as one system that understands the broader state of work and helps move it forward continuously.

## The best systems will feel calm

As models become more capable, intelligence itself becomes less differentiating. The real challenge becomes judgment: knowing when to act, when to interrupt, when to wait, and what actually deserves attention.

The best systems will not feel like software constantly demanding interaction. They will feel calm. They will quietly handle coordination in the background, surface what matters at the right time, and give people back the attention required to do their best work.

That is the kind of software we believe the next decade will be built around.

---

## The Operator

URL: https://orchid.ai/blog/the-operator
Published: 2026-05-25
Topic: Essay

Every piece of software ever built assumed you'd be the one doing the work. The operator era was a tax on getting what you want. It's over.

Every piece of software ever built has the same assumption baked into it.

You.

You are the one who will read, decide, type, click, drag, search, compare, copy, paste, and send. The software just waits. A machine full of potential, doing absolutely nothing until you show up and start operating it.

We called this "using a computer." But look at what it actually is. You are the labor. The software is the supervisor. It presents you with a screen. You perform. It presents you with another screen. You perform again. It never does anything. You do everything.

The entire history of personal computing is a history of making humans better operators. Faster keyboards. Better interfaces. Fewer clicks. Every generation is the same promise: we made it easier for you to do the work.

Nobody ever stopped to ask why you were doing the work.

## The contract

Somewhere along the way, we signed a deal we never read.

The deal was this: technology gives you capability, and in exchange, you give it your time. Your attention. Your effort. Every single interaction.

Want to go somewhere? You operate the airline's booking software. Want to talk to someone? You operate the email client. Want to find something? You operate the search engine. Want to organize your day? You operate the calendar. You are always the operator.

And the strange thing is, none of these tasks are the point. The flight is the point. The conversation is the point. The information is the point. The day is the point. But you never get to just have the point. You have to perform a sequence of operations to reach it.

That sequence is not work. It's a tax. A tax on getting what you want.

We've been paying it so long we forgot it was there.

## The gap

There is always a distance between wanting something done and having it done.

Last month one of our customers hit a milestone with us. I wanted to send them something. That was the whole thought. Send them a gift. Five seconds of intent.

What followed was 45 minutes. Figuring out what to send. Browsing three sites. Wondering if they drink coffee or tea. Looking up their office address in an old email thread. Copying it into a shipping form. Picking a card. Writing a note that didn't sound generic. Entering my payment info. Checking the delivery date.

The decision to be thoughtful took a moment. The execution of that thoughtfulness took the better part of an hour. And the worst part is, I almost didn't do it. Not because I didn't want to. Because the gap between wanting to and actually doing it was wide enough that it nearly killed the impulse.

But between the want and the outcome, there is always a gap. And for the entire history of computing, you have been the thing that fills it. You are the one who translates intent into action, click by click, screen by screen, form by form.

You are the middleware.

Every productivity tool ever built accepted this as a given. Of course there's a gap. Of course you fill it. We just made the gap smaller. Fewer clicks. Faster loads. Better design. The gap shrank, but it never closed. Because closing it would mean the software doesn't need you in the middle anymore.

Nobody built for that. Because until now, it wasn't possible.

## The workaround

Every task you perform that isn't a decision is a workaround.

Typing a search query, writing a status update, scheduling a meeting, comparing options, looking up an address, filling out a form, drafting a message in someone else's expected tone. All workarounds. Things you do because the machine couldn't. It was never "your job." It was a limitation of the technology, and you were the patch.

## The oldest solution

Some people figured this out a long time ago. They didn't wait for better software. They made a hire.

An executive assistant. A chief of staff. Someone who reads the inbox before they do, moves the calendar without being asked, sends the follow-up before it slips. Someone who fills the gap so they never have to.

The people who have this don't think about it. That's the whole point. Things just happen. They say what they want and the outcome appears. No tabs. No forms. No procedure. The gap doesn't exist for them because someone else closes it.

This has always been the answer. Not a better tool. A person who handles the work so you don't have to operate anything.

The problem was never the model. The model is perfect. The problem was that it cost $5,000 a month, took months to set up, and scaled to exactly one person.

So most people went without. And kept being the operator.

## The end of operating

The next era is not a better interface. It's no interface. Not in some abstract, sci-fi way. In a plain way.

You say what you want. It happens.

Not because the software got smarter at showing you screens. Because the software stopped needing you to operate it at all.

The flight gets booked, the email gets sent, the calendar sorts itself out, the gift ships, the information surfaces. Not because you clicked through seven screens. Because you didn't have to.

This is not a feature. It's not a product category. It's the end of a relationship that never made sense. The relationship where the most sophisticated technology on earth still needed a human to click "next."

## Why we built Orchid

We built Orchid because we were tired of being the operator ourselves.

We spent our mornings chasing the same things everyone chases. The inbox. The calendar. The follow-ups. The busywork that feels productive but isn't. We kept building tools to manage it better, and then we stopped and asked a different question: why are we managing it at all?

Orchid is the hire most people never get to make. Not a tool you operate. Not a dashboard you check. A hire. Something that shows up before you do, handles what it can, and surfaces only the things that actually need you.

We don't think people need better software. We think they need to stop being the infrastructure of their own lives.

The operator era is over.

---

If you've ever felt like the work between you and the thing you actually wanted was wider than it should be, we built Orchid for you. Stop being the middleware. [Try Orchid](https://orchid.ai) and let it run the gap for you 🌸

---

## Introducing Voice Notes: You don't think in sentences

URL: https://orchid.ai/blog/introducing-voice-notes
Published: 2026-05-20
Topic: Product

Voice notes are live on Orchid SMS. Hold the mic, talk, and the work gets done while you're already doing something else.

Your brain doesn't think in sentences. It thinks in flashes. *Remind me. Email her back. Pull that up. What did Jake say?*

You've been translating those flashes into typing for decades. The friction has become so constant, so taken for granted, that it doesn't even register as friction anymore. It just feels like work.

It isn't. The work is the decision. The typing is the tax.

Today we're shipping the way Orchid was always supposed to feel. You hold the mic, you talk, and the rest happens.

## How it works

{{VIDEO}}

Open SMS, hold the mic, say what you want. Orchid transcribes it instantly, figures out what you meant, and either does the thing or asks you a clarifying question.

*"Email Sarah back and tell her Tuesday works."*

*"Pull the invoice from Stripe for that refund."*

*"Remind me to check in with the design team Thursday."*

*"What did Jake say about the contract last week?"*

You don't write any of that. You just say it.

## The original interface

Voice is older than every keyboard, every screen, every app. It's how humans have been delegating work to each other since the first chief gave orders to the first lieutenant. It's how the most powerful people in the world still operate. Founders walking the halls. Executives in the back of cars. Lawyers between depositions.

They don't type their own emails. They tell someone what to do.

The reason you've been typing isn't that typing is better. It's that nothing on the other end was smart enough to listen yet. That part isn't true anymore.

## When the work happens

The biggest thing voice notes change isn't *how* you work. It's *when*.

Typing requires a desk, a keyboard, a moment of stillness. So all the small admin work piles up and waits for you to sit down. It steals the start of your morning. It eats your evening. It guards every transition between meetings like a tollbooth.

Voice removes the desk.

The walk to the coffee shop. The drive home. The five minutes before the next call. Those used to be empty. Now they're work happening in the background of your life. You finish a walk and a dozen things are already done.

That isn't a productivity boost. It's a different relationship with your day.

## What you can talk into existence

Voice notes don't replace what Orchid does. They just remove the writing.

Everything you can do on SMS, you can now do on the move:

- Brain dump your morning walk. Arrive at the office with the day already queued.
- Talk out the awkward email you've been avoiding for three days. Approve it.
- Capture an idea before it slips away mid-step. Memory holds it for you.
- Negotiate the calendar with three people without ever opening it.
- Pull context from threads from months ago, without a search bar.
- Trigger the workflows you've already built, without remembering their names.

And you can ramble. Change your mind mid-sentence. Repeat yourself, contradict yourself, trail off. Orchid figures out what you actually meant and confirms before doing anything that matters.

That's the part most voice products miss. Voice is only useful when it doesn't ask you to speak like a robot. It has to let you speak like a human.

## Where this is going

Today, you talk and Orchid handles the small stuff.

Soon, the surface widens. The voice note stops being an instruction and starts being a thought you shared with a partner who already knew what you wanted. The reach grows. The approval moment shrinks.

We can already feel that shift. Walking and talking to Orchid feels less like dictation and more like thinking out loud with someone who has been with you since the start. Someone who already knows the deal you're chasing, the lawyer you keep meaning to call, the founder waiting on a reply.

This is the interface we've been pointing at since the beginning. You stop operating software. You start having a relationship with it.

## The last keyboard

There will be a generation that doesn't learn to type. They'll talk to their work and watch it happen. The keyboard will be the thing in the corner of the office that nobody uses, like a fax machine.

We're not there yet. But every voice note you send is a small bet on that future. And every time it works, it's a little harder to imagine going back.

If you're already on Orchid, hold the mic next time you reach for SMS. If you're not yet, [get access](https://cal.com/team/orchid/meet) and we'll get you set up. As always, if you want to riff on what we're building, reach out: [nizzy@orchid.ai](mailto:nizzy@orchid.ai) 🤍🌸

---

## The True Cost of Hiring an Executive Assistant in 2026

URL: https://orchid.ai/blog/the-true-cost-of-hiring-an-executive-assistant-in-2026
Published: 2026-05-18
Topic: Essay

The salary is the easy number. Benefits, ramp, turnover, and the hours you spend managing them are the rest. Here is what an EA actually costs in 2026, and how to think about the math.

Ask anyone who has hired an executive assistant what it costs, and most will quote a salary. That number is rarely the real one.

The full cost of an EA in 2026 is a layered thing. Salary is one slice. Benefits, recruiting, training, equipment, turnover, and the hours you spend managing them are the rest. By the time you add it all up, the true number is often two to three times what people first quote.

This is a breakdown of what an EA actually costs in 2026, where the hidden costs live, and how to think about it against the alternatives.

## The salary, by tier

EA salaries in the US sit in three rough bands, with national averages around $80,000 to $95,000 for the role.

- **Junior EA**: $50,000 to $75,000 per year. Calendar coordination, inbox triage, travel logistics. Usually 0 to 2 years of experience. Best for principals whose admin needs are mostly volume, not judgment.
- **Mid-level EA**: $75,000 to $130,000 per year. The default hire for most operators. Strong calendar judgment, can draft client communication in your voice with light supervision, runs travel and gifting end to end.
- **Senior EA**: $120,000 to $160,000 per year. Operates as a force multiplier. Top of the band overlaps with the lower end of chief of staff comp. Can attend meetings on your behalf, hold institutional memory across hundreds of relationships, and run cross-functional projects.

Chief of staff is a related but distinct role, with US average base around $218,000 and a typical range of $164,000 to $304,000 according to recent Glassdoor data. Startups tend to pay lower base, in the $125,000 to $170,000 range, with equity making up the difference. If your principal really needs a chief of staff, that is a separate hire and a separate budget conversation.

In high-cost cities like New York, San Francisco, and London, add 15 to 25 percent to any of these bands. In specialized fields like law, finance, or executive search, add another 10 to 20 percent on top.

## The hidden costs

The sticker price is the part everyone quotes. The hidden costs are where the real number gets built.

- **Benefits and payroll taxes**: roughly 30 to 40 percent on top of base salary. Health insurance, retirement match, payroll tax, paid time off, sick leave. According to BLS data, benefits make up about 30 percent of total compensation for private industry workers in late 2025. A $100,000 base becomes $130,000 to $140,000 fully loaded.
- **Recruiting**: $15,000 to $40,000 if you use a placement firm. Contingency firms typically charge 20 to 30 percent of first-year base salary, with retained search firms going higher. Even in-house, expect 30 to 60 hours of your team's time over the search.
- **Onboarding and ramp**: a great EA takes three to six months to reach full output. During that window you are paying the salary while spending your own hours teaching them your preferences, voice, and judgment. Conservatively, that is another 80 to 120 hours of your time. Multiply by your hourly rate to see the real figure.
- **Equipment and software**: laptop, phone, subscriptions, expense card. Usually $3,000 to $5,000 in year one and $1,500 to $2,500 each year after.
- **Turnover**: EA tenure tends to run three to five years on average, longer than many admin roles because of the close relationship with the principal. But when they do leave, you pay for the search again, the ramp again, and the institutional memory walks out the door. Industry research puts the true cost of replacing a professional employee at 50 to 150 percent of their annual salary, with the higher end common for roles where context loss is significant.

Add it all up. A mid-level EA on a $100,000 base actually costs closer to $145,000 to $170,000 per year fully loaded once you account for benefits, equipment, and amortized recruiting and ramp costs. A senior EA on a $150,000 base lands between $215,000 and $245,000 once everything is counted.

## The time tax most executives forget

The cost most operators miss entirely is the one they pay in their own time after the EA is hired.

A good EA needs direction. They need feedback, course corrections, weekly priorities, context on relationships, and access to your judgment when something does not fit the playbook. Most operators spend three to five hours a week managing their EA. At a $400 hourly rate, that is another $60,000 to $100,000 a year of your own time, layered on top of the EA's compensation.

That is not a complaint. It is the deal. A great EA earns those hours back many times over. But it belongs in the math.

## What you actually get

Not all EAs are created equal, and the gap between an average EA and a great one is enormous.

A great EA reads your inbox faster than you can, writes in your voice indistinguishably, knows which meetings you would hate before you do, and holds institutional memory of every relationship in your network. They protect your time aggressively, fix problems quietly, and make you look more organized than you are. Twelve months in, they know your preferences better than you can articulate them.

An average EA executes the tasks you assign. They schedule what you ask them to schedule. They reply to what you forward. They are useful, but they do not compound.

The price of a great EA is higher, the search is longer, and the retention is harder. Most operators try and fail to hire one two or three times before they get the right person.

## The alternatives, and what they actually cost

The full-time US-based EA is the gold standard and the most expensive option. There are three real alternatives, each with a different shape.

- **Offshore EA**: $1,200 to $4,000 per month, usually based in the Philippines or Latin America. Lower cost, but time zone friction, language nuance, and limited institutional context typically mean you get a fraction of what a US-based EA delivers. Premium offshore services with heavy training and management overhead, like Athena, Persona, and Pearl Talent, can run $2,500 to $5,000 per month.
- **EA agency or fractional service**: $2,500 to $5,000 per month for US-based agency support. A trained EA who works across multiple clients. Faster to onboard, no benefits or recruiting overhead, but you share their attention with other principals.
- **Executive assistant service that runs on software**: $50 to $400 per month, depending on the depth of the service. The newest category. The good ones handle inbox drafting, calendar coordination, meeting prep, and routine follow-ups, and they run on your approval so you stay in control while the work moves in the background.

## The math by role

The same EA cost lands very differently depending on what your hour is worth.

- **Founder of a $5M revenue company**: roughly $300 to $600 per effective hour. A fully loaded EA at $170,000 pays for itself in 340 recovered hours per year. That is about seven hours a week. Most founders recover more than that in the first month.
- **Solo law firm partner**: the average US lawyer hourly rate is $349 per Clio's 2025 Legal Trends Report, with solo practitioners typically billing $150 to $350 per hour and specialists charging $500 to $800 or more. Three to four recovered billable hours per week, at a $300 rate, covers a $150,000 EA.
- **Real estate agent doing $200,000 in GCI**: one additional closed deal per quarter from faster follow-up covers most of an entire year of EA support. The 80 percent first-responder advantage in speed-to-lead means most agents are leaving more than that on the table.
- **VC or investment partner**: dealflow is a relationship business. The deals you lose to slow follow-up are usually invisible until they show up in someone else's portfolio.

In every case, the math works for the right person. The reason most people who could afford an EA do not have one is not cost. It is the search, the management, the ramp, and the risk of getting it wrong.

## Where Orchid fits

We built Orchid for the people who looked at this math and got stuck on the hiring part. The salary made sense. The fully loaded cost made sense. They just did not want to run another search, another onboarding, another three-month ramp, another performance review.

Orchid handles inbox triage, drafts replies in your voice, runs calendar coordination, preps you before every meeting, sends the follow-ups you forgot, and lives in Messages so there is no new app to babysit. It works on approval, in your voice, quietly in the background.

For most operators, Orchid sits at a fraction of the cost of even an offshore EA, with none of the recruiting or onboarding overhead. It is not for everyone. If your needs are complex, fully delegated, or deeply relationship-heavy, you will still want a great human EA. If your needs look like the eighty percent that an EA actually does day to day, Orchid covers it.

## A question worth sitting with

If you priced out your last twelve months of admin work at your real hourly rate, what would the number be?

And what would change in your week if that number went somewhere else?

---

## How to Run Your Week Like You Have a World-Class EA (Without Hiring One)

URL: https://orchid.ai/blog/run-your-week-like-you-have-a-world-class-ea
Published: 2026-05-15
Topic: Playbook

Your time is a company asset. The real question isn't 'how do I get more done?' It's 'how do I stop spending my most expensive hours on work an EA should own?' A playbook.

If your time is a company asset, the real question isn't "How do I get more done?"

It's "How do I stop spending my most expensive hours on work an EA should own?"

This post is a playbook for that. It's how to run your week _as if_ you had a world-class executive assistant, even if you haven't hired one yet.

The punchline: you'll see how much of your week is EA work, not CEO/partner work, and why having a proper EA setup (whether human, firm, or EA-powered service) is not a luxury, it is infrastructure.

## Step 1: Put a price on your time

You can't design a smart week if you don't know what an hour of your time is worth.

- Take your annual income or the annual value you're responsible for (e.g. profit, revenue, deals closed).
- Divide by roughly 2,000–2,500 working hours per year.

If you're at $600k/year, that's roughly $250–300/hour. If you're at $1.5M/year, that's closer to $600–700/hour.

Now ask: "How many hours last week did I spend on things I'd _never_ pay someone $300/hour to do?"

Most executives find at least 10–20 hours of their week fits that description once they're honest.

## Step 2: Audit your week like an EA would

For one week, track your time with only two buckets:

- **High-leverage**: decisions, deals, strategy, key relationships, top hires
- **EA work**: scheduling, rescheduling, chasing people, travel logistics, prepping basic context, inbox triage, status pings

You don't need a fancy tool. Go back through your calendar and inbox and reconstruct the last 5 working days. Rough is fine.

Questions to ask yourself:

- How many hours were you in meetings you should never have attended?
- How many hours disappeared into calendar coordination, email back-and-forth, or "just five minutes" tasks?
- How many important things _didn't_ happen because you didn't have the time or mental space?

Write down the total EA-work hours. Multiply by your hourly rate from Step 1. That number is the invisible EA line item you're already paying, whether or not you've hired anyone.

## Step 3: Decide what only you can do

A world-class EA doesn't just "do tasks." They make sure _you_ are doing the right work and _they_ are doing everything else.

Make three lists:

1. **Only I can do**
   - Final decisions on strategy, key hires, investments
   - High-stakes external conversations (fundraising, biggest clients, board)
   - The few things where your unique judgment or reputation is non-negotiable

2. **EA should own end-to-end**
   - Calendar design and protection
   - Scheduling and rescheduling
   - Gathering docs and context before important meetings
   - Drafting routine communications and chasing follow-ups
   - Travel and logistics

3. **Shared / transitional**
   - Things you're doing now that an EA could own with the right guardrails (e.g., some internal updates, some vendor comms, some approvals)

Be aggressive. Everything that doesn't _clearly_ belong to "Only I can do" is a candidate for EA ownership.

## Step 4: Redesign your calendar around leverage

Without an EA, your calendar is usually a random list of other people's priorities. With real executive support, it becomes a reflection of your actual job.

Use these rules to redesign it:

- **Block your "money time" first.** Reserve 2–3 chunks per week for deep work on revenue-moving or company-shaping work. Treat these as unbreakable.
- **Cluster similar meetings.** Put external calls together, internal check-ins together, etc. Context switching is where a lot of energy dies.
- **Create hard rules for what gets a meeting.** For example: no 30-minute calls that don't have a clear decision or outcome; asynchronous first for low-stakes topics.
- **Protect recovery.** Leave real buffers; don't run your brain at 100% utilization.

If you had a world-class EA, this is what they'd be doing for you behind the scenes. You're just role-playing that system yourself for now.

## Step 5: Build simple "EA scripts" for your future self

Until you have an EA (or EA-style service) to run this, you can at least codify your preferences like one.

Create a short EA brief for yourself:

- How you want your mornings to start
- What types of meetings you _never_ want on your calendar
- What information you need before any important conversation
- Who has priority access to your time and who doesn't

Then turn those into simple rules:

- "If a meeting doesn't have a clear agenda and output, it gets pushed back or moved to async."
- "Investor/client calls get a minimum 15 minutes of prep time protected beforehand."
- "No more than X hours of meetings in a single day unless it's planned."

This is the raw material a real EA or EA firm would use to run your week.

## Step 6: Decide whether to keep DIY'ing this

If you do Steps 1–5 properly, two things will happen:

1. You'll get immediate relief. Your week will feel less chaotic.
2. You'll see, in black and white, that you're running an EA system manually on hours that are too expensive to waste.

That's the moment where a real EA setup becomes obvious:

- A full-time human EA if you have the budget, volume, and appetite to hire and manage.
- An EA firm or EA-powered service if you want the outcomes without the headcount and overhead.

Either way, the math doesn't lie: once your effective hourly rate crosses a certain threshold, continuing to self-manage your week is one of the most expensive decisions you can make.

## Where Orchid fits in

Orchid exists for people who look at this playbook and think: "Yes, this is exactly how I want my week to run, and no, I don't want to be the one doing it."

World-class executive assistant support so your time goes into decisions, deals, and strategy, not juggling details. Orchid keeps your week pointed at the work only you can do and clears the mental clutter around it. You operate like someone with a top EA watching your back.

If you had to guess, based on last week, how many hours were truly EA work for you, and what does that number become when you multiply by your real hourly rate?

---

## This Is Not What You Were Hired For

URL: https://orchid.ai/blog/this-is-not-what-you-were-hired-for
Published: 2026-03-24
Topic: Essay

The productivity industry made management lighter. It never made it gone. There's a different model, one where you stop managing and start deciding.

There is an associate at a law firm somewhere right now who spent three years in law school and two more grinding through junior work to get good at one thing: the law.

Today, she spent 40 minutes following up on a signature. Twenty minutes scheduling a deposition. Fifteen minutes drafting a routine correspondence that says nothing a template couldn't say.

None of that is the law. All of it is management.

And here is the part that makes it absurd: the firm billed $400 an hour for it.

Not because her time isn't valuable. Because there was no other model. Someone had to chase the signature. Someone had to find the time. Someone had to send the email. And she was the someone.

This is not a law firm problem. It is everyone's problem. It just happens to be most visible when the math is written in billing rates.

## The trap nobody names

Every productivity tool starts from the same premise: you are the manager. You are the one who reads, decides, routes, responds. The tool just helps you do that faster or with less friction.

Get a better email client. Write replies faster. Hire an EA. Find a tool that manages your calendar. Each one saves time. Each one is a genuine improvement. And yet, at the end of it all, you are still managing something.

Just less of it.

That is the trap. The entire productivity industry is built on making management feel lighter. Lighter is not the same as gone.

The question nobody asks is not "how do I manage this better?" It is "why am I managing this at all?"

## Being delegated to

We talk a lot about delegation. Handing things off. Getting things off your plate. But delegation still requires you to initiate. You have to know what to delegate. You have to find someone to delegate to. You have to explain the context. You are still doing work.

There is a different model. One where you don't delegate outward. Where work flows toward you, already prepared, already handled, waiting for your approval.

You are not the manager. You are the decision-maker. The final step. The one who says yes or no.

Everything else happened without you.

That is not delegation. That is being delegated to. And it is a fundamentally different relationship with your work.

## The model that already exists

Some people already live this way. Partners with assistants. Executives with chiefs of staff. Someone who arrived before them, read everything, handled the routine, and left only the decisions on the desk.

The assistant didn't ask for permission to do that work. They used judgment. They acted. They prepared. And they surfaced the thing that actually needed the principal's attention.

The bar for "needs you" is strict. A real decision. A deadline. Someone blocked. A high-stakes call. Everything else gets handled. Silently. In the background. Without interrupting the day.

Most people don't have that. Most people are still the someone who has to chase the signature.

## Most routine work doesn't need a human to manage it

It needs a human to approve it. Those are not the same thing.

One requires your full attention: read the thread, understand the context, decide what to do, do it. The other requires a moment of judgment: does this look right? Yes. Done.

Collapsing that distinction is what the productivity industry has never done. Every tool assumes you need to be present for the work. The better model assumes you only need to be present for the decision.

That shift sounds small. It isn't. It changes what your day looks like entirely.

## Back to the associate

Now imagine she opens her laptop in the morning and there are three things waiting.

The signature was chased. A response came in overnight. It is ready to review.

The deposition is scheduled. Three options were proposed, opposing counsel confirmed the second. The calendar hold is drafted.

The correspondence went out. The reply is already summarized.

She tapped approve on each one in under two minutes. Then she went and did the law.

That is not a better tool. That is a different world. The work happened. She let it happen.

Most people don't have access to that world. Orchid is what it feels like when they do.

---

## Orchid Beta: Your Proactive Agent for Work

URL: https://orchid.ai/blog/orchid-beta-is-here
Published: 2026-03-15
Topic: Product

Orchid connects to your inbox, calendar, and other tools. It reads what's coming in, prepares the next action, and waits for you to approve.

Today we're opening up Orchid to our first users.

Orchid connects to your tools, understands your work, and handles things before you even think about them. You just approve.

### Not faster. Gone.

Every product right now makes the same promise: _we'll help you do your work faster_. Better search, smarter summaries, assisted drafts. That's a 1-2x improvement. You're still the one doing the work, just with a slightly better tool in your hand.

Orchid is something different. The work doesn't reach you at all.

You open your laptop in the morning. 44 things have already been handled. 3 need you. That's not a faster inbox. That's a categorically different relationship with your work.

The distinction matters because speed isn't the problem. The problem is that you're carrying it all in your head. _Did Sarah get back about the contract? Should I follow up with that lead? When was that investor meeting?_ That background noise runs all day, even when you're not at your desk. Orchid eliminates it. Not by helping you process it faster, but by making sure it's already taken care of before the thought even crosses your mind.

### Delegation, not efficiency

There's a reason the best executives hire a chief of staff. It's not because they can't do the work. It's because doing the work keeps them from doing _their_ work.

Orchid is that shift, but available to everyone. It reads what's happening across your work, understands your relationships, your priorities, how you communicate, and starts acting on your behalf. You stay in the loop on the decisions that matter. Everything else just gets done.

The difference is how it makes you feel. You stop feeling like someone who manages a business and start feeling like someone who runs one. That's not a productivity improvement. That's an identity shift.

We believe the future of work is humans and agents collaborating together. Right now, you approve everything. Over time, as trust builds, the agent takes on more. The surface of what it handles keeps growing. You stay in control, and the rest just gets done.

### Who this is for

We built Orchid for ourselves as founders, but the more we've tested it, the more obvious it is that this isn't just a founder thing. If you spend too much of your day on the _work around the work_, Orchid is for you.

Sales teams buried in follow-ups. Real estate agents juggling dozens of client threads. Operators coordinating across teams. Agency owners managing client communication. Recruiters keeping candidate pipelines moving. Really anyone who feels like the repetitive, operational side of their job is taking over the part they're actually good at.

### Where we are

This is a beta. It's real, it works, and we use it every day. But we're also first-time founders shipping something new and we know there's a lot to improve. We'd rather get it in your hands now and build with your feedback than wait for perfect.

If you want to try it, connect your Google account and Orchid starts learning in minutes. We're opening this up to our first 100 users.

If you want to talk about what we're building or how it could fit into your workflow, reach out directly: [nizzy@orchid.ai](mailto:nizzy@orchid.ai)

---

## The End of Inbox Zero and What Replaces It

URL: https://orchid.ai/blog/the-end-of-inbox-zero
Published: 2026-02-12
Topic: Essay

Inbox Zero made you the processor. The future is an agent that treats email as context, not a destination.

We've been sold a lie for over a decade.

Inbox Zero. The idea that if you just process every email, archive it, label it, snooze it, delete it. You'd finally feel in control. You'd be productive. You'd be free.

Except you're not free. You're just managing.

You spend 15 minutes, maybe an hour, every single day sorting through emails. Reading things that don't matter. Deciding what to reply to. Flagging things you'll forget about. And when you finally hit zero, you feel good for about five minutes before the next batch rolls in.

Inbox Zero doesn't eliminate the work. It just makes you the processor.

And here's the kicker. Even the people who are great at it, the ones with perfect systems and zero unread, they're still spending time that could be spent elsewhere. Whether it's an hour or fifteen minutes, you are still the one processing. You are still the bottleneck.

## The false belief

Inbox Zero assumes that the problem is organization. That if you just had the right system, the right labels, the right workflow. Email would stop being a burden.

But that's not the problem. The problem is that **you're doing the work at all.**

Think about it. Why are you the one reading every email? Why are you the one deciding what's important? Why are you the one drafting replies, scheduling meetings, processing refund requests, creating tickets?

You're not being productive. You're being a router. A human middleware between your inbox and your actual work.

We talked to dozens of founders and operators while building Orchid. Not a single one said "I wish I could organize my email better." What they actually said was: _"What needs my attention? I want to see everything that needs me, all in one place."_ One founder told us: _"I don't want another inbox. I want to stop emails from cluttering my life and only be notified on what matters."_

People don't want a better inbox. They don't want an inbox at all.

Inbox Zero optimizes the wrong thing. It optimizes how fast you can process. It doesn't ask: should you be processing at all?

## Why do we even manage inboxes?

Because we had no other choice.

Email was designed in a world where humans were the only ones who could read, understand context, and take action. So we built systems around that constraint. Filters. Labels. Priority inboxes. Stars. Snooze buttons. All of it designed to help _you_ be a better processor.

But we're not in that world anymore.

Agents can read. Agents can understand context. Agents can take action.

So why are we still sitting in front of the inbox?

## Email isn't the product. It's the context.

This is the part everyone gets wrong. Every new email app tries to be a better client. Better UI. Better search. Better AI summaries. They're all still building inboxes.

But email was never the thing. Email is where all the signal lives.

Your inbox knows who you talk to. How often. What about. It knows your writing style, your relationships, your priorities. It knows when you're busy, when you're responsive, when you go dark. It knows about your deals, your customers, your team dynamics, your deadlines.

No other tool has this much context about you and your work. Not your calendar. Not your project tracker. Not your CRM. Your inbox is the single richest source of information about who you are professionally and what you need to do.

And yet every product built on top of email treats it as the end destination. A thing to look at. A thing to manage.

That's the wrong frame entirely.

**Email is not the product. Email is the source of context that makes everything else possible.**

One of the things we learned early: context isn't a feature, it's the foundation. Without it, you're just another automation tool. With it, you have something that actually understands your world.

## What replaces the inbox

Not a better client. Not a prettier UI. Not AI summaries on top of the same inbox.

What replaces the inbox is an agent that treats email as input. It reads everything. It builds an understanding of your work, your relationships, your priorities. And then it uses that understanding to actually do things.

A customer emails asking for a refund. The agent doesn't just surface the email. It already knows this person, knows their account, pulls the invoice from Stripe, and prepares the refund for you to approve.

Someone wants to schedule a meeting. The agent doesn't show you the email. It checks your calendar, finds the right time, and drafts the invite. You approve it.

A bug report comes in. The agent reads the thread, extracts the details, creates the Linear ticket, and drafts a reply to the customer. One click.

The inbox never enters the picture. You never open it. You never manage it. The agent consumed the context and did the work.

That's not a better inbox. That's email as a context layer for an agent that works for you.

## Connecting to everything

Email gives you context. But work happens across tools.

Your calendar, your issue tracker, your payment processor, your CRM. The real power comes when the agent can move across all of them. And the email context is what makes it possible. The agent knows _why_ it should create that ticket, _who_ the customer is, _what_ was promised, because it read the thread.

Right now, MCP is the best way to connect these tools. You give the agent access through a standard protocol and it can act across all of them. My belief is that agent-to-agent communication will get better over time. We'll move from agents calling APIs to agents talking to each other. Your email agent coordinating with your calendar agent, your project management agent. But we don't need to wait for that future. MCP works now, and it works well enough to build something real.

One founder we interviewed put it perfectly: _"I want an agent that has full context and can run workflows."_ Another wanted to _"go on a walk, have a conversation with the agent, and just get things done."_

They're not describing a better inbox. They're describing a system where email is the context and the agent is the worker.

## From processor to supervisor

This is the fundamental shift.

The old model: you sit in front of your inbox, you read, you decide, you act. You are the processor.

The new model: the agent reads your email, understands the context, connects to your tools, prepares the work. You review and approve. You are the supervisor.

**Email is input, not product.** One click can trigger actions across Stripe, Linear, Calendar, and email simultaneously. That's not reading email. That's delegating work.

And "needs you" has to mean something real. Not everything that lands in your inbox needs you. We define it strictly: a decision is required, there's a deadline, someone is blocked, there's a high-risk situation, or you explicitly asked to be told about something. Five gates. If it doesn't pass one of them, it shouldn't interrupt your day.

## Not autonomous. Not yet.

I want to be clear about something. We're not building an agent that sends emails on your behalf without asking. Could that happen one day? Probably. But that's not where trust is right now.

What we're building is an agent that **prepares** actions for you. It drafts the reply. It sets up the refund. It creates the ticket. It finds the calendar slot. And then it shows you what it's about to do and waits for your approval.

One click. That's it.

You go from being the person who reads, thinks, decides, and executes. To the person who reviews and approves. And the agent gets better the more you use it, because every approval teaches it your writing style, what you care about, who matters to you, and how you make decisions.

The through-line we heard across every conversation with users: **people want to feel in control while doing less work.** They want an agent that knows everything, does the routine stuff silently, and only interrupts them for decisions that actually need them.

## Inbox Zero is dead

It served its purpose. It gave us a framework when we had no other option. But the world changed.

The next iteration of email isn't a client. It's not an app you open and manage. It's a source of context that powers an agent to do the work for you. You stop being the processor. You become the person who watches your agent handle it.

That's what we're building with Orchid.

Not a better inbox. Not an inbox at all.

---

If this resonates, I'd love to hear from you. Reach out directly: [nizzy@orchid.ai](mailto:nizzy@orchid.ai)

---

## Rethinking AI In Applications

URL: https://orchid.ai/blog/a-new-generation-of-ai-applications
Published: 2025-09-24
Topic: Notes

Most apps copy the chat-box model. The future is invisible, effortless AI.

AI is having the wave of the decade in both consumer and business software. If you're even somewhat online, it feels like a new AI powered app drops every day, with no end in sight. And while new tools are extremely exciting, it got me thinking about the bigger picture of AI powered apps: are we really implementing AI the right way?

In all honesty, this question first came to mind while we were working on our first email client: Zero. We wanted to implement AI (like every founder does) with the end goal of saving our users time in their inbox. However, it felt like every AI feature we added, the harder and less enjoyable it was to use the app.

You might then be asking, "_how does adding AI into an app make it harder to use_?". This came down to one issue: the way users interacted with the AI was suboptimal and quite unenjoyable. It added friction instead of removing it.

So what were we doing wrong? We were thinking of AI implementation like a developer instead of an industry expert.

### The problem: "Cursor for [insert industry]"

Right now, tons of new and existing apps have the "AI-powered" label. However, that typically translates to a simple chat box that spits out text, usually in the form sidebar / chat interface. In my opinion, this form of AI interaction can help in some contexts, but is it really the best pattern for user workflows? I'd argue not.

So, if AI isn't optimal in this chat interface, why does everyone do it? The answer is actually quite simple: People building AI applications (developers) saw the success of Cursor (an AI powered code editor that uses a sidebar to communicate with the agent) and figured that it's the most optimal way of using AI in every other app.

That is exactly what we did with the initial versions of Zero. We simply added a chat interface into our app, an AI composer section, a few tool calls to add labels, and called it a day. There was no true time savings or productivity in what we built. Our users got no benefit.

This was a 2D way of thinking: _copy a pattern that works (Cursor and ChatGPT) and hope it works everywhere else_. I also saw this first hand in the X25 YC batch. Lots of very talented founders and developers who had amazing ideas were building with the exact same AI pattern as Cursor did and hoping it would work. The end result? Lots of "Cursor for [insert industry]" instead of tools that actually meet people where they are.

### The real goal: fewer steps to the outcome

If AI is going to play a role in everyday software, it should remove steps, not add them. A chat box is fine, but it should be a secondary option. It should be a piece of the pie, not the entire pie. The way to interact with AI should almost feel effortless and invisible, almost like it's reading the users mind.

Here's what this looks like in Orchid:

- AI drafts email in your voice when a new email comes in, all you have to do is approve
- AI automatically shows you what you do and don't care about via previous actions
- Feature that lets you organize your inbox with a click of a button via AI

These actions don't make you "think in prompts." They surface likely intents and put them one tap away. Less "operate the tool." More "the tool already knows what you probably want." This approach cuts friction, lowers brain load, and truly saves time without forcing you to learn a new interface.

### Our current stance

Sidebar chat can be helpful, but it often turns into a detour. You switch context, type an instruction, wait for an answer, then copy whatever it says back into your actual task or wait for the implementation by the agent. It might be cool to you but it's not that helpful to your end user.

We're not spiking the football and the solution is still evolving, we are first-time founders with a lot to learn. But our direction is clear: make AI that closes the distance between intent and outcome. Make actions obvious, contextual, and fast. Keep the chat input, but treat it like a power tool, not the default.

Do this well and AI gets quieter and more useful. It feels like the product understands what you're trying to do and just helps you get there faster. this requires more creativity but it's well worth it to the end user.

If you want to jam on what we're building at Orchid and why we think this model actually serves people better, ping me. Happy to share more: [nizzy@orchid.ai](mailto:nizzy@orchid.ai)