Private beta · Your team's AI headquarters
Everyone on your team can ship.
Product managers, designers and support describe what they want, or just click on it in your app. AI teammates build it on your real product and show it running. Engineers review and ship.
- Fits your GitHub, GitLab, CI and Jira
- Your AI plans and machines, no markup
- Set up once for the whole team
Walkthrough: a designer comments on a button in the live preview, the Coder Bot changes the codebase and runs the tests, the preview reloads, an engineer approves from their phone, CI fails and autopilot fixes it, and the change is merged.
Ask from where your team already works
- Slack
- Jira
- Linear
- GitHub
- GitLab
- Sentry
- Your running app
For everyone on the team
Ideas stop waiting for the next sprint.
The people closest to customers ask in plain words. Engineers get finished, tested work to review.
Engineers stay in control
Nothing ships without review.
Teach each AI teammate your codebase, conventions and rules once. Every change is tested, explained and reversible.
- Review finished work
- A plain-language summary, the diff, the running preview and the checks in one place. Approve or ask for changes from your desk or your phone.
- Roles that match your team
- Anyone can ask, comment and preview. Only engineers approve, merge and change how AI teammates work.
- Isolated by default
- Every task gets its own copy of your repositories and its own running services. Nothing touches main until it is merged.
- Reversible
- Every step is a snapshot. Fork from any point, try another approach, or undo.
Make Export the primary action
Asked by Maya, Design · built by Coder · 6 minutes
Export CSV is now the primary action on Reports. Filter and Share stay secondary. No API, data or permission changes.
- Types
- Lint
- Tests · 48 passed
- Preview
| Permission | Requester | Engineer |
|---|---|---|
| Ask, comment and preview | Yes | Yes |
| Approve and merge | No | Yes |
| Configure AI teammates | No | Yes |
Gets better with every task
It asks for what it's missing.
Your AI teammate notices what it lacks and asks right in the conversation. Approve with one click; the whole team benefits.
- A tool
- It suggests the integration it needs, from GitHub and Sentry to your own CLI.
- An access key
- With step-by-step help. The key goes straight to your runner, never into the chat or the model.
- A team convention
- Learned from corrections and review comments, written into its instructions.
Bring your own
Your AI. Your machines. Your bill.
Use the Claude and ChatGPT plans your team already pays for, your API keys, or open models. No AI markup. Run it on our cloud, your servers, or entirely inside your network.
- Our cloudHosted runners start when work arrives and stop when idle.
- Your serversRun the runner on your own machines, with your own logins.
- Inside your networkHost the whole thing. Nothing leaves your infrastructure.
Every action and dollar visible
Know what it did, and what it cost.
Cost per person, project and model; budgets; a full record of everything the AI did.
Compared with
Not another tool for engineers to type faster.
- Lovable
- Great for new prototypes.
- AIHQ works on the product you already have.
- v0, Builder.io
- Edit your frontend.
- AIHQ changes the whole product, frontend to database, and follows it through to merge.
- Copilot, Cursor
- Help engineers type faster.
- AIHQ lets the whole team ship.
- Devin
- An AI engineer for engineers.
- AIHQ is AI teammates for the whole team, on any model, on your infrastructure.
Pricing
One price for the whole team.
Pay for the people who start work, not everyone who reviews it. AI usage runs on your own keys, with no markup. AIHQ is in private beta: we onboard teams one at a time.
FAQ
Questions engineers ask first.
Can non-engineers break production?
No. Nothing ships without engineer review. Requesters ask, comment and preview; only engineers approve, merge and change how AI teammates work.
Is my code safe?
Code, credentials and AI logins stay on the runner: in our cloud, on your servers, or inside your network. Secrets go straight to the runner and never appear in the chat or reach the model. Every action is recorded.
Which AI models?
Claude, ChatGPT and open models through compatible endpoints such as OpenRouter or Ollama. Pick per task; instructions, integrations and services stay the same.
Does it work with our stack?
Any stack, frontend to database. Every task gets an isolated copy of your repositories and your real services running, so changes are checked against the whole product.
What happens when it gets stuck?
It asks, in the conversation, for exactly what it needs. Failing CI and review comments are fixed automatically a few times in a row, then it hands back to you.
Can I use it from my phone?
Yes. Follow progress, answer questions and approve work from your phone; AIHQ installs to your home screen like an app.
What does it cost?
AIHQ is in private beta; request access to join. Free on your own runners. Teams pay $299 a month for up to 10 active users, including a hosted runner; see Pricing. AI usage runs on your own API keys or Codex login, so you pay model providers directly and see every dollar in AIHQ.
Engineering is no longer the bottleneck.
Give your whole team AI teammates that work on your real product. Engineers stay in control.