Connectors
Connect once; stay current without thinking about it. Syncs run in the background, ingest only what changed, and prune what was removed.
How syncing works
- Instant response, background work. Clicking Sync returns immediately; ingestion continues in the background and the connector page shows live progress.
- Incremental by content. Every synced item is fingerprinted. Re-syncing skips anything already ingested and unchanged; a 500-email mailbox re-syncs in seconds.
- Pruning. Items deleted at the source are removed from your knowledge base on the next sync.
- Resilient. One bad document doesn't fail the batch; it's skipped and reported while the rest ingest.
- Browsable. Every connector page lists its synced documents. Click any item (an email, a page, an event) to see exactly what was ingested.
Google (Gmail · Calendar · Drive)
Google connectors use OAuth: you sign in with Google and grant read-only scopes.
- ConnectOn the Connectors page, choose the Google tool and click Connect. You'll be sent through Google's consent screen and back.
- SyncClick Sync. It returns instantly; progress streams on the connector page.
- AskEmails, events, and files are now part of your answer surface, cited like any other source.
Gmail scope
By default Gmail syncs recent, meaningful mail: the last 180 days, excluding promotions, social, spam, and trash, with a per-sync cap. This way first sync is fast and the index stays high-signal. The query and cap are adjustable per connector.
Calendar & recurring events
Recurring meetings are collapsed into one document per series (with its schedule and details), not one per occurrence. Your weekly stand-up is context, not 52 duplicates.
access_denied, the app's OAuth consent
screen is in testing mode and your Google account isn't listed as a test user. Add the
account under Google Cloud Console → APIs & Services → OAuth consent screen → Test users,
or complete Google's verification for production use.
Notion
- Create an integrationIn Notion, create an internal integration and copy its secret (starts with
ntn_). - Share pages with itThis is the step everyone misses: the integration can only see pages explicitly shared with it. Open the top-level page(s) you want indexed → ⋯ → Connections → add your integration. Child pages inherit access.
- Connect & syncPaste the secret in the connector and sync. All shared pages are ingested with full pagination, so large workspaces are fetched completely, not just the first page of results.
Slack, Confluence, Salesforce, HubSpot
| Connector | What syncs | Credential |
|---|---|---|
| Slack | Channel messages. The decisions buried in threads become searchable. | Bot token |
| Confluence | Wiki spaces and pages. | Site URL + API token |
| Salesforce | Accounts, contacts, opportunities, queryable in plain English. | OAuth / token |
| HubSpot | Contacts, companies, deals. | Private-app token |
Databases (PostgreSQL · MySQL · MongoDB)
Point Vectropic at a database with a connection string and its schema becomes part of the question surface: "how many orders shipped late last month?" compiles to a guarded query against your own database.
- Read-only by design. Use a read-only database user; queries are additionally guarded and row-capped on our side.
- DSN format:
postgresql://user:pass@host:5432/dbname(equivalents for MySQL and MongoDB).
Coming soon
Meeting transcripts from Google Meet, Zoom, and Microsoft Teams. Connectors that aren't configured for your workspace appear in the gallery marked Coming soon.