The Company Agent Platform
San Francisco, California
- Tweets 1,099
- Following 199
- Followers 5,359
- Likes 217
Your company, your agent.
A few beliefs about building a universal company agent:
1. You should own it, and it should feel like yours.
2. One agent surface. Not a swarm of fake coworkers with personalities and unique capabilities.
3. It should integrate with anything that has an API, not just MCP.
4. Every piece should be swappable for the best in the world at that piece. A full-stack AI platform, not a wrapper.
5. It should run anywhere, from on premises to fully managed. No lock-in.
6. It shouldn't be a framework or an unneeded abstraction. It should be just code you can run, own, and control.
adapt.com
We believe every company will have a central company agent.
Today a company's intelligence is scattered across its tools and the few people who know where everything lives. The company agent changes that.
At Adapt, our job is to build the infrastructure for your company agent.
It's time to build.
1
4
12
1,243
More details around why we built this here:
adapt.com/blog/branded-slack…
2
427
This week in Adapt was supposed to go out Friday. It's Monday. Whoops, so it goes.
• App version history: every app keeps a running history of its versions, and you can switch between them right in the UI instead of guessing what changed.
• Personalized prompt suggestions: up to three integration suggestions based on your work function and your org's actual usage, so you hit a real "aha" faster than a generic starter list gets you there.
• Smarter channel setup: when Adapt joins a Slack channel, it now prompts members to set up SMART trigger rules on the spot, so channels get the right reply behavior from day one.
• Chrome Extension v0.2.0: remote tool calls, so actions you start in the browser reach the same tools Adapt uses everywhere else.
Plus per-page tab titles, native tables in Slack replies, organization logos, cleaner fullscreen apps, and four modal implementations consolidated into one.
adapt.com/blog/changelog-202…
8
490
Introducing user-based knowledge.
Shared knowledge is the secret sauce for how an organization gets work done. But the people who use Adapt want it to work the way they work, with their own style and norms.
User-based knowledge unlocks things like:
• Emails that actually sound like you
• Code written the way you'd write it
• First drafts of blog posts in your own voice
Watch the video to see how it works, and let us know what you think.
2
7
936
Three launches this week, and a theme: Adapt gets more personal and more hands-on.
Here's everything we shipped:
• Browser actions: with Chrome extension v0.2.0, Adapt can reread a page as it changes, inspect a specific form, fill fields, and click controls in the tab you started from. Clicks pause for your approval by default.
• Personal knowledge: Adapt now separates your preferences, writing style, and templates from shared company knowledge, automatically. Yours follows you; the org's stays the org's.
• Organization logos: admins can upload a company logo that shows up in the workspace menu, so a shared Adapt feels like part of your company, not ours.
Plus quality-of-life upgrades: cleaner navigation, better mobile settings, voice drafts that persist, readable tables on mobile, and active chats that survive a refresh.
1
4
520
Take action in your browser, knowledge that's actually personal, and a workspace that looks like your company.
We build in the open and write these roundups by hand, every week.
Read the full roundup 👇
adapt.com/blog/changelog-202…
1
230
This already exists, and it's exactly what we built Adapt to be.
• Model-agnostic by design: every task routes to the best model, no single-vendor tag
• Feels like a teammate: it lives in Slack, you just tag it in
• Powerful enough for engineers: every task spins up its own sandbox, a real computer where the model writes code, runs it, and ships the result
The trick is simple: you understand the workflow and the steps, capture them as a reusable skill, then let Adapt go to work as your integrated coworker.
Wrote up exactly how to build this below.
anyone have a model agnostic (so no claude tag) slack agent that can do stuff like this? devin automations not quite there
want it to really feel like a teammate - but also the dev infra/coding has to be good too.
rn I feel like my best hope is to wait a month or two for devin to get this right
1
5
610
How to build a natural-language agent, the same way you onboard a new hire: do the work together, capture it as a skill, then flip on proactive mode.
adapt.com/blog/how-to-build-…
2
545
Factories, loops, and more. Read all about it from our very own @Jumballaya
Loops, factories and graphs oh my! The AI dev world acts like these showed up last week. They didn't. Same stuff we've done in software for decades, handed to an agent.
Start at the bottom. An LLM predicts tokens in a loop until it hits a stop. Wrap a program around it that acts on the output and feeds it back, and you've got an agent. That's the agentic loop. No magic.
Loop the loops and you get a factory. The shape those steps make, prerequisites and parallel bits, is a graph. And no, graphs aren't replacing loops. They wrap them. Pull the loop out of each node and you've got workflow automation with an LLM taped on, the thing we already got bored of.
The truth is if you fan out twenty loops with no gate and you burn 20x the tokens. If the loop was wrong, you paid to be wrong twenty times. I call that review debt, and the only people that benefit are the model providers.
So the real work is the gate. Small scope, quit early, a graph that says what's worth doing, a human who still approves the merge. My janitor's put up 66 PRs that way, every one merged.
Wrote it up: adapt.com/blog/agentic-loop
7
397
Whew. This team ships.
Six headline launches this week plus a couple of quality-of-life upgrades, and honestly, it was just a normal week for us. Here's everything:
• Policies + Groups: set natural-language access policies for any integration, scoped to groups or named people. e.g. Engineering gets full read/write on Linear while everyone else stays read-only.
• Messages: text Adapt over iMessage, RCS, and SMS, with the same history, integrations, knowledge, and skills you use everywhere else.
• Premium research: the search agent now pulls specialized, structured data (company + people data, web traffic, market data, product catalogs, and more), cross-checks it against the open web, and answers with citations. No setup.
• Gated PR reviews: "Adapt Review" runs as a required GitHub check. 3/5+ passes; anything lower blocks the merge.
• Bring your own Shopify app: connect Shopify via your own OAuth app, choose the read scopes, and let Adapt handle token refresh + webhooks.
• Workflows: real, reusable examples we run internally, including Code Janitor, Docs & Changelog, and SEO Agent. adapt.com/workflows
• Chat persistence: unsent composer text now survives reloads and chat switches.
• Connection health alerts: broken integrations get flagged instead of failing silently.
1
1
5
534
That's group-based access policies, text-your-agent over iMessage, deeper research, gated PR reviews, bring-your-own Shopify, and real team workflows, all in one week.
We build in the open and write these roundups by hand, every single week.
Read the full roundup 👇
adapt.com/blog/changelog-202…
176
Your integrated coworker now has a phone number.
Adapt now works in Messages: iMessage on iPhone, RCS on Android.
Connect your phone like any integration, then get real work done with your company's context and tools.
We help you get real work done, wherever you want to do it.
5
3
16
3,447
Read all about it with the release note:
adapt.com/changelog/messagin…
2
226
Welcome @adaptRyanMiyaha to Adapt as a Product Engineer.
Ryan joins with a deep developer-centric background with his experience building tools used by millions of developers around the globe and is already helping build the knowledge layer that makes our integrated coworker smarter every day. He shipped his first PR on day one and a meaningful feature week one. As we expect all new engineers to do. 🚢
adapt.com/blog/welcoming-rya…
2
1
8
374
The meaningful feature he shipped is a small, but important one that you may not have even noticed you needed. Now chats are cached automatically and when you come back to a draft chat, it will persist.
adapt.com/changelog/chat-per…
152
Research shouldn’t be an arduous hero's journey where you open 30 tabs and hunt, sprawl, and peck to get insights.
Adapt now combines the open web with specialized data to research companies, markets, competitors, products, and more.
Ask once. Understand immediately.
Available now on all paid Adapt plans.
3
2
9
729
We just hit the APIs directly.
Check out our blog post covering some of our sandbox infrastructure adapt.com/blog/orchestrating…
2
160