The to-do list for your agents

TL;DR: Koma is now your agents' favorite to-do list. Connect the agents you already use, bring in the work they do, and put the decisions you need to make in one place. Welcome to the command center for you and your agents. The source of truth for everything that got done, is getting done, and needs to get done.

Why we're here

Last week we launched Koma - the to-do list that does itself. Almost immediately we heard, "can we connect our agents?" We had already been doing this internally, so this made us super excited to double down.

We're all using a lot of agents at once. They're doing more work around the clock, but it's still up to us to organize it. Some of that work lives in a chat, results live in another, and context gets lost along the way.

We believe in agency over agents. Your work shouldn't vanish into a chat thread. It should be yours to see and to steer.

Koma is the command center for you and your agents.

One home for the work

Connect your agents to Koma via MCP, giving it a shared place to read and update the work. Bring in all the agents you already use: Claude, Codex, OpenClaw, Cursor, Gemini, Muse, Instinct, even the ones you built yourself.

Just like you, they can create and update tasks, search your list, read your projects. Whether your agent runs in an app or on your machine, its work has the same home. You don't need to replace the agent or move all your work into a new chat.

Koma holds the context for that work: the plan, the progress, the result, any major decision. When an agent checks back, it picks up from that record instead of asking you to reconstruct everything in an entire conversation.

Plus it brings the work back to you: morning briefings, reminders when things are due, a heads-up before a meeting. You answer in the task in Koma, and your answer is there when your agent checks back in. The agent does the work. You make the call.

Two ways to use it

One list for everything that needs your attention. One user runs a hardware store and a small consultancy. Twice a day their scheduled agent reads their inbox, calendar, meeting recaps, and CRM they've given access to. It doesn't reply. It adds the task in Koma they need to handle, with the source and deadlines - a bank reply, reviews to prepare for, drafted emails to approve. They open Koma, read the briefing, and work through the list. On the agent's next run, it checks Koma for what changed before starting more work.

The agent does the work. You make the call.

Another user runs a company with a coding agent beside them all day. The agent prepares replies, reconciles invoices, and gathers compliance evidence in files on their machine. The only thing it can’t do is decide: approve a draft, choose an option, sign a document. So that’s what it files in Koma: the context, what it already did, and a link to the work. They answer in the task. Their answer is there for the next session, so the agent doesn’t ask twice.

Memory doesn't live in a chat

Without Koma, each agent builds its own context. You repeat yourself. You chase down what happened. You rebuild the plan every morning.

Now your context has one home. Tasks are the source of truth for what needs to happen and how it was handled. The agent writes its progress there, and your decisions stay there. The next session starts from the written record, not from scratch.

And it compounds. The more your agents work through Koma, the less you repeat, the less you chase, and the faster everything moves.

Setting it up

You’re probably already running an agent. Maybe Claude with connections to Gmail, Granola, and your CRM. Maybe a routine that checks your inbox every morning. Maybe a coding agent open all day.

Connect Koma to it all. Then tell the agent what to do: read Koma first, do its work, and file what’s waiting on you before it finishes. That’s it. Your tasks start appearing in Koma, with their status, their context, and the decisions waiting on you.

Where this goes

We’re already seeing it - adoption has been amazing. Some of you haven't missed a day since launch, close dozens of tasks in a day, and have Claude, Cursor or ChatGPT filing work into Koma around the clock.

Today: all your agents now have one home. No more chasing. No more lost context. One single view of all your work and your agents. Tomorrow: all of this for you, your agents, and your entire team.

If you run agents, connect one and tell us how you’re using it. If you’ve built one, we’d love to see what it does with a to-do list.

Welcome to your agents’ command center. One place to see the work, keep its context, and decide what happens next.