AI Chat

The hosted chat requires an authenticated account with current Terms, Privacy, service-data authorization, and age confirmation. Anonymous prompt dispatch is disabled.

Models and tools

The deployed model catalog can include providers configured through Vercel AI Gateway and other configured routes. The model selector, account entitlement, provider credentials, and runtime availability jointly determine whether a request can run. Provider names in documentation are not availability promises.

The free plan currently receives 50 credits on initial registration or transition back from paid, and access to a selected model set. Paid self-service checkout is suspended by default. Tool, web-search, deep-research, sandbox, and memory behavior depends on feature flags, credentials, plan enforcement, and the active Life runtime.

Conversations and content

Authenticated conversations and messages may be persisted to the application database and sent to the selected AI/tool providers as described in the Privacy Policy and Subprocessors page. AI output can be inaccurate and must be independently reviewed for consequential use.

The interface renders streaming Markdown, code, math, diagrams, and tables when the response contains those formats. Conversation sharing and organization features remain subject to their route-level authorization controls.

External API

Use POST /api/v1/chat/completions for the OpenAI-compatible external API. The in-app /api/chat endpoint is an internal UI-message SSE contract. See the Chat API for the supported request shape and current limitations.