Home / FAQ
Didn't find your answer? Message us on Telegram @DmitryLunevsky or on LinkedIn.
No. AI Council connects to Claude Code, Codex and other MCP-capable agents. Your team keeps working in the same setup as today.
A rules file works well in a small project. AI Council ties knowledge to the code, keeps it separate per branch, checks it for freshness and delivers it to the agent on its own. Detailed comparison
Python, TypeScript and JavaScript, Java, Kotlin, C#, C, C++, PHP.
From analyzed tickets in GitHub Issues and Redmine, project documents, project chat answers, and entries that agents and people add while they work.
When the code under an entry changes, a similar entry appears or a branch is merged, the entry goes to review. The model adds a label with a verbatim quote from the source, and the entry text stays as it was. A human sees the labels in the log and can undo any of them.
When work starts, the agent gets every project-level rule in full, and after that they are repeated on a rotation. Rules for a specific file are picked by their bindings.
Yes. The platform is installed on your server with a license key. The model side runs through Claude Code CLI, which can connect to an open-source model on your servers. Air-gapped installs are part of the Enterprise plan.
When you self-host, your code and knowledge base stay on your server.
Telemetry never includes user content, prompts or branch names. More on security
Startup, for teams of up to 10 users: $30 per user per month. Standard, for 10 to 50 users: $20 per user per month. Both plans run in the cloud or on your own server. For more than 50 users or an air-gapped install, we prepare an Enterprise quote. See pricing
It's a monthly subscription billed per user. You subscribe online, and checkout and billing are handled by Paddle.
Yes, at any time. There is no long-term contract. Refund policy
Yes. Add or remove seats as your team changes.
No. You pay the model provider directly, under your own subscription or API key. AI Council sends the agent only the context the task needs, so you spend fewer tokens.
If you decide not to continue, we'll help you export your knowledge base and rules.