AI assistant & agent

Configuring who runs the models, the assistant, the build agent, the concierge, integrations, workflows and MCP.

• AI configurationWho runs the models for your organization — your own provider keys, or ours. The two modes, what self hosted asks for, what managed costs, and who is allowed to change it.
• The AI assistantWhere the assistant shows up across your rooms, what it reads, and what has to be configured before it will answer.
• The build agentConnect a repository to an AI agent room and turn stories into pull requests, from first prompt to review.
• Reusable & personal agentsSave an agent's job once, connect it to as many rooms as you like, and let each room decide what it may actually touch.
• Agent integrationsConnect WordPress, Slack, calendars, Stripe, CRMs and helpdesks as agent tools, and control when the agent may write to them.
• Workflows, actions & skillsAutomate recurring agent work with workflows, extend the agent with custom HTTP actions, and shape its behavior with skills and knowledge.
• MCP serversExtend the agent with tools from any MCP server, under two-step human approval and organization-wide oversight.
• Board MCP serverConnect Claude or ChatGPT to one Kanban board over MCP — read the columns, write stories, move cards, leave comments, and run a dev/QA loop between agents.
• AI usage & the token allowanceThe token allowance caps what the platform pays for, not what you may use. What counts, how to read the bar, what happens when you hit it, and what became of Openbook's credits.
• The conciergeDescribe the team and the work, and the concierge builds the spaces, rooms, boards and stories — then hands the stories to an agent room.