AI Chat & MCP¶

UC-AI-01 Start chat session¶

As a member, I want to start a new AI chat session, so that I can ask questions and get answers within my organisation context.

Acceptance: Session created; messages stored; JWT org enforced.

As a member, I want to reopen a past session, so that I can continue work without losing context.

As a member, I want to choose the model provider (OpenAI / Gemini endpoints), so that I use the best model for the task.

As a member, I want to run vision/OCR on an image, so that handwritten or scanned content becomes text for the assistant.

Constraint: Subject to plan vision quota.

As a member, I want to generate images from prompts, so that I can illustrate ideas in the playground.

As the platform, I want each chat request to record token spend, so that balances and billing stay accurate.

As an AI agent (via chat), I want to list, read, and create documents through MCP, so that answers reference real user files.

As an AI agent, I want to fetch the latest tablet screenshot, so that the model can reason about the user's current notes visually.

As a member, I want to send a tablet command from chat, so that the assigned device performs an action allowed by the service.