Integration · ai

Anthropic Claude.

The AI assistant runs on Anthropic Claude — Sonnet 4.6 for day-to-day, Opus 4.7 for deeper reasoning. We don't train models. Conversations log in your tenant, scoped to your organization at the database, and never cross-organization. Anthropic's API receives only the context we explicitly assemble per turn.

What It Does.

Streaming chat with conversation persistence
Tool calling scoped to your workspace data
Per-org model selection + monthly cost cap
Prompt caching for repeated context
Audit log: who asked what, which model, which data touched

Technical Anchor

Streaming chat at /api/v1/ai/chat (claude-sonnet-4-6 / claude-opus-4-7 via @anthropic-ai/sdk)

Modules That Depend On It.

AI assistant Advancing Interactive Proposals Safety Stack

Anthropic Claude · FAQ

Is our data used to train Claude?

No. Anthropic's API is zero-data-retention by default for our usage tier — your prompts and completions aren't stored beyond the request lifecycle on Anthropic's side, and they don't train on them.

Can we self-host the model?

Not today. Festival tier customers can BYO an Anthropic API key if they have a separate ZDR contract or specific compliance needs.

Other Integrations.

Stripe Connect Supabase Twilio Resend Google Apple Microsoft Sentry

ATLVS Is Open.

Free for small teams. All integrations included.