Connect AI assistants to your users' data

Ask a question and get an AI-generated answer from the user's connected accounts (Gmail, Slack, Calendar). Returns answers with optional source citations.

Semantic search across the user's connected services. Returns matching messages grouped by conversation without AI summarization.

List calendar events from connected calendar accounts (e.g., Google Calendar). Returns event details including attendees, location, and meeting links.

List meetings with AI-generated summaries and action items from connected meeting platforms (Google Meet, Zoom, Teams).

List the user's connected integrations and their current status. Useful for checking what data sources are available before running queries.

Returns integration ID, provider name (gmail, slack, google_calendar, etc.), and connection status (connected, disconnected, expired, error, pending).

List notes, decisions, and observations already attached to the authenticated user context.

Write a note into the authenticated user context so future agents can reuse decisions, observations, or session state.

Save a meeting transcript or summary into the authenticated user context from another meeting MCP, a transcript export, or manual notes.

Reversibly archive a pushed meeting by opaque id or external_id. Archived meetings are hidden from meeting/search/query results and can be restored by re-pushing the same external_id.

Reversibly archive a note by opaque id or external_id. Archived notes are hidden from notes/search/query results and can be restored by re-pushing the same external_id.

Create a user-scoped outcome goal with a watch scope anchored by entities, a seed query, or keywords.

List outcome goals with optional lifecycle and health filters.

Get a goal with its latest status snapshot, evaluator run, risk signals, next actions, and cited evidence.

Tail goal lifecycle transitions as an append-only feed with cursor pagination and a composite scan watermark.

Queue a fresh asynchronous evaluator run for an active goal.

Attach a manual note to a goal so future evaluations can cite it.

Manually confirm or override a goal lifecycle and/or health with a reason.

Acknowledge that a goal's silence is expected and suppress the silence-driven health escalation until a horizon (or until real activity arrives). Does not change lifecycle or suppress non-silence risks.

Clear the active acknowledgment on a goal and resume normal silence monitoring immediately. Idempotent: succeeds even if there is no active acknowledgment.

Draft the next move on a quiet goal as a ready-to-copy follow-up email, grounded in the goal's real thread evidence. Read-only: nothing is sent, drafted with a provider, or persisted.

Register a promise, delegation, or handoff that will remain unresolved after this session, so Attrove watches it until the expected signal arrives and escalates to a human if it silently never does. Underspecified registrations are kept as low-confidence suggestions that never escalate.

Read back open commitments (watching/at_risk, optionally suggested) at the start of a session, soonest-due first with cursor pagination. A to-do ledger, not memory: resolved history is never returned.

Record what actually happened to a watched commitment: satisfied (the signal arrived), silent_drop (it never came), dismissed (not a real commitment), or at_risk (flag without closing). Terminal states are immutable.