Step 1
Deploy an agent
Start from a ready-made template — like a Code Reviewer or a Researcher — or spin up your own with a custom persona and instructions.
Hand a task to an autonomous AI agent and live your life while it works. Give it a mission in plain language — it runs through visible steps, draws on your AI Brain for context, and reports back when the job is done.
Research Agent
Running autonomously
How it works
No prompt engineering, no babysitting. Four steps from idea to a result you can use.
Step 1
Start from a ready-made template — like a Code Reviewer or a Researcher — or spin up your own with a custom persona and instructions.
Step 2
Tell it what you want in plain language. No prompt engineering, no rigid commands — just describe the outcome you are after.
Step 3
The agent works through visible steps, drawing on your AI Brain for the context it needs — while you get on with your day.
Step 4
Get a clean result in report mode, or let it act on your behalf in managed mode. Either way, the work comes back done.
Deploy a proven agent in one tap, or craft your own: write the persona, set the instructions, and point it at your data. Every template is a starting line, not a cage — tune it until it works the way you do.
Browse templatesCode Reviewer
Reads a diff, flags issues, and explains what it would change.
Researcher
Gathers sources on a topic and distills what actually matters.
Inbox Triage
Sorts a pile of items and surfaces only the ones that need you.
Writer
Drafts a plan, brief, or post from your notes in your own voice.
Your own agent
Custom persona + instructions, built by you.
Run more in parallel
Higher tiers run more agents at once — and let each one go deeper, with more steps and longer autonomous runtime.
Free
1×the agent limit
Where everyone starts
Deploy an agent and watch it work, end to end — no credit card.
Pro
5×the agent limit
More agents, longer runs
Run more in parallel on deeper, longer missions.
Max
25×the agent limit
Your full workforce
Run a whole team in parallel, each one going deeper with longer autonomous runtime.
The agent does the legwork and brings back clean findings for you to review. Nothing changes until you say so.
Give it the reins and the agent acts on your behalf — filing notes to your Brain, organizing data, and finishing the job.
Example missions
A few ways people put agents to work. Say it once, in your own words, and let the agent take it from there.
“Summarize my saved articles into a weekly brief.”
A tidy digest of themes and takeaways, saved to your Brain.
“Review this code diff and flag anything risky.”
A list of issues with the reasoning behind each one.
“Research electric bikes under $1,500 and save the key facts.”
Sourced findings filed into your AI Brain as Knowledge.
“Draft a launch plan from my scattered notes.”
A structured first draft you can refine in seconds.
“Triage my reading list and surface what is worth my time.”
The few items that matter, ranked — the rest set aside.
“Pull together this month and tell me where my money went.”
A plain-language breakdown of spending by category.
Deploy your first agent free — no credit card. Upgrade to run more in parallel, up to 25× the agent limit, whenever you are ready.