File a request.
Watch it ship.
Capybari is a fully managed AI engineering organisation. A project manager plans your request. Specialists build it on isolated branches. A code reviewer and a security engineer sign it off. Then it deploys to your own servers — with health checks, automatic rollback, and someone on call when it goes wrong.
Regression pass on card and wallet payments
Implement 3-D Secure challenge step
Challenge modal and retry states
Threat review of the payment flow
Assess current checkout against SCA rules
Migration: store authentication outcomes
Teams shipping with Capybari
The gap
You do not need more code.
You need the rest of the team.
Code generation stopped being the bottleneck a while ago. What small teams still cannot staff is everything that happens around the code — the planning, the review, the security pass, the deploy, and the person who gets up when it breaks.
Not a copilot
An organisation, not an assistant
Fourteen specialist roles, each with a written contract defining what it owns and where it must stop. They hand work to each other the way a real team does.
- A Project Manager — reads your request and writes the plan — in plain English, with owners and dependencies.
- Governance roles — an architect, a code reviewer and a security engineer who can reject work, not just produce it.
- Capability substitution — if a role is missing from your team, the work is reassigned to whoever can cover it — and the substitution is recorded, never silent.
Isolation
Every task gets its own branch
Tasks run in parallel, each in its own git worktree on its own branch. No agent can see another's half-finished work, and none of it reaches your default branch until the gates pass.
- Dependency-aware — a task that depends on another simply waits. The graph is visible on the board.
- Reversible by construction — an abandoned task is a deleted branch. There is nothing to unpick.
- Your git history stays legible — one branch per task, one report per branch.
T-02 task/t-02-auth-migration → .worktrees/t-02
T-04 task/t-04-sca-challenge → .worktrees/t-04
T-05 task/t-05-challenge-modal → .worktrees/t-05
Gates
The part everyone else skips
Any role can be marked as requiring human review. When it is, everything downstream stops until you approve — at every autonomy level, without exception.
- Structured reports — every task ends with a summary, the files it changed, and its risks, assumptions and blockers.
- Agents that ask — when a specialist hits genuine ambiguity it stops and asks you a question instead of guessing.
- Nothing merges quietly — you see the diff, the reviewer's verdict and the security note before anything ships.
Operations
It deploys. And it stays.
Connect a server over SSH and give Capybari a deploy command, a health check and a rollback command. It merges, ships, checks, and puts things back if the check fails.
- Automatic rollback — a failed health check restores the last commit known to be healthy. Typically inside a minute.
- Investigations — a failed deploy opens an investigation. An agent reads the logs, finds the cause and writes it up.
- The fix becomes a request — the investigation opens the follow-up work itself, linked back to the incident.
Merged 4 branches, no conflicts
Deployed to prod-web-1 over SSH
Health check failed — /healthz returned 503 twice
Rolled back to a3f91c2 — the last commit known healthy
Investigation INV-11 opened automatically
How it works
Request to production, in five moves
The same pipeline every time. What changes is how much of it you approve by hand.
You file a request
A sentence is enough. "Card payments fail for European customers." Type, priority, done.
It gets planned
The Project Manager reads your codebase and writes a task plan with owners and dependencies. You approve it or edit it.
Specialists build
Tasks dispatch in parallel, each on its own branch, each ending in a structured report.
Gates and merge
The reviewer and the security engineer sign off. You approve anything marked for review. Branches merge.
Deploy and watch
It ships to your server, runs your health check, and rolls back on its own if the check fails.
By the numbers
What is actually in the box
Product facts, not projections. Every figure below describes something shipped today.
We publish no customer counts, no "hours saved" and no velocity multipliers until we can show you the method behind them. See our roadmap for what is coming next.
The org chart
Fourteen roles. One board.
Add the specialists a project needs and leave out the ones it doesn't. Work planned for a missing role is reassigned to whoever can cover it.
Coordinator
Coordinator
Governance
Governance
Governance
Worker
Worker
Worker
Worker
Worker
Worker
Worker
Worker
Worker
Every role is governed by a Markdown contract you can read and edit. See how the team works.
Built in
The things you would otherwise build yourself
Autonomy you dial up
Start at Manual and approve everything. Move to Supervised when the plans stop surprising you. Reach Autonomous when you trust the gates. Most teams take about three weeks.
Investigations
When production misbehaves, an agent is dispatched with read-only access. It reads logs and diagnostics, writes up the cause, and opens the request that fixes it.
Opened by the failed deployment of REQ-142
Read journalctl -u tilltap-api on prod-web-1
Cause found — SCA_PROVIDER_KEY is not set on the production host
Opened REQ-143 to add the variable to the deploy configuration
Your repositories
Connect any git remote. Capybari works on branches inside your repo — not on a copy in ours.
Design previews
Front-end work generates a shareable preview link so stakeholders can look before anything merges.
Editable role contracts
Every role's behaviour is a Markdown document. Tighten it, loosen it, or write your own.
Live run streaming
Watch a task work in real time — the commands it runs and the output it gets back.
Talk to the manager
Chat with the Project Manager about a project. It can turn the conversation into a request.
Rollback that has run
Not a checkbox in a settings page. A command you supply, executed the moment a health check fails.
In practice
What changes when the gates are someone else's job
We moved from Manual to Autonomous over about a month. The thing that convinced me wasn't a feature — it was watching it roll back a bad deploy at 2am and open the investigation before I woke up.
I am not an engineer. I file requests in the same language I would use with a contractor, and I get a plan back that I can actually read before anything happens.
Our reviewer catches the things our juniors used to ship. Having a security pass on every request stopped being a luxury the week it found a card number in our logs.
Read the full stories in our case studies.
Pricing
Pay for work done, not seats filled
An agent run is one task executed end to end by one specialist. That is what we meter, because that is what produces something. Viewers are free.
Free
One project, real gates, no card.
1 project · 100 agent runs a month
Everything you need to see it work:
- 3 core roles — manager, full stack, reviewer
- 1 repository, 1 project
- Manual autonomy
- 1 run at a time
- Community support
Starter
For one team shipping one product.
1,000 agent runs a month
Everything in Free, plus:
- 8 specialist roles
- 3 projects, unlimited repositories
- Supervised autonomy
- 1 deployment target with rollback
- 3 concurrent runs
- Email support, one working day
Pro
For teams that want it to run itself.
5,000 agent runs a month
Everything in Starter, plus:
- All 14 roles plus custom roles
- Unlimited projects
- Autonomous autonomy
- Unlimited deployment targets
- Production investigations
- 10 concurrent runs
- Priority support, four hours
Business
For agencies and multi-client work.
20,000 agent runs a month
Everything in Pro, plus:
- Bring your own model keys
- Per-client project isolation
- Audit log export
- 25 concurrent runs
- Named contact and onboarding
Going over your included runs bills at the overage rate — it never blocks a deploy mid-pipeline. Full pricing and comparison.
Objections
The questions you are actually asking
Does Capybari need access to my production servers?
Only if you want it to deploy. A deployment target is an SSH host you define, with a deploy command, a health check and a rollback command that you write. Plenty of teams run Capybari for planning and building only and never connect a server at all.
What stops an agent breaking my main branch?
Every task runs on its own branch inside its own git worktree. Nothing touches your default branch until the review gates you configured have passed and the merge succeeds.
Any role can be marked as requiring human review. When it is, dependent work stops until you approve — at every autonomy level, with no exceptions.
What happens when a deployment breaks production?
Your health check fails, your rollback command runs, and an investigation opens automatically. An agent with read-only access reads the logs and diagnostics, writes up the cause, and opens a follow-up request containing the fix.
Write access to a production host is a separate, explicit grant. It is never the default.
Do I have to let it run on its own?
No, and we would rather you didn't at first. Projects start at Manual, where nothing runs until you press the button. Supervised starts planning on its own and dispatches work you have approved. Autonomous lets healthy work ship without you.
Review gates and genuine escalations reach a human at every level.
Is my code used to train models?
No. Your repositories are cloned to run your work and nothing more. We do not train on your code, your prompts or your reports. The security page sets out where your data lives and who can reach it.
What if the plan it writes is wrong?
Reject it, edit it, or say what it missed and have it planned again. Nothing is dispatched from a plan you have not approved unless you have deliberately turned that approval off.
Can it work on an existing, messy codebase?
That is the case we built for. Agents read your conventions before they plan, and the role contracts instruct every specialist to make the smallest safe change and to reuse existing patterns rather than introduce new ones.
Are you ready for my enterprise?
Almost certainly not, and we say so on the pricing page with the specific list of what we are missing. If you need SSO at identity-provider grade, a completed SOC 2 audit or a contractual uptime SLA today, we are not your platform yet.
Still circling something? Ask us directly — a person replies.
Start with one request
Connect a repository, file one thing that has been on your list for a month, and watch what the plan looks like. Manual mode. Nothing runs without you.
No card required · Free plan forever · Export everything, any time