← all help · Knowledge

// knowledge

KNOWLEDGE OVERVIEW

How BandBrain ingests, indexes, and reasons over your tour documents.

The Knowledge page (sidebar → Knowledge) shows every document BandBrain has access to for your tour. Day sheets, advance docs, contracts, contact lists, settlement notes — anything you've connected.

How files get in

Three paths:

1. Connected cloud storage (Google Drive / Dropbox)

Set up at /dashboard/connectors. After connecting, you pick a Drive folder or Dropbox folder. BandBrain watches it — files added/edited/removed in that folder get auto-synced.

See Connecting Google Drive and Connecting Dropbox.

2. Email forwards / attachments

Your band has a dedicated inbox at <yourband>@bandbrain.ai (Channels → Email shows it). Anything sent to that address gets ingested:

  • Attachments become resources (just like Drive files)
  • Email body text is parsed for structured facts (load-in time, doors, hotel confirmation)
  • Links to Drive/Dropbox files get auto-followed and ingested

This is how Master Tour day-sheet emails land in BandBrain — your booking agent emails them to your BandBrain address, attachments get sucked in, day_facts get extracted, the agent can answer questions about the show within minutes.

What happens after ingest

For every new file:

  1. Discovery — Anthropic Sonnet reads the file and figures out what it is (day sheet, advance doc, contacts list, contract, etc.), what its tabs/columns mean, and writes a structured schema description.
  2. Indexing — file content is chunked + embedded via Gemini, stored in pgvector for semantic search.
  3. Day-facts extraction — Sonnet pulls structured per-date facts (city, venue, hotel, schedule, parking, etc.) into a day_facts table. This is what find_today reads.

This all happens in background jobs, typically within a couple minutes of the file landing.

What the agent can do with it

  • Semantic search — search_knowledge retrieves passages relevant to a question
  • Tab/column reading — read_resource_tab reads specific rows of a structured sheet
  • Image OCR — read_resource_image reads scanned day sheets / photos
  • Day-fact lookup — find_today queries the structured facts table for time-bound questions

Freshness

When a file in your connected folder changes, BandBrain re-runs discovery + indexing + extraction automatically. There's no "refresh" button anywhere in the UI — freshness is automatic by design. See Why there's no refresh button for the why.

Coverage

The Coverage page shows a matrix of what the agent can correctly answer for upcoming show days. Useful for spotting gaps (e.g. "no hotel for next Tuesday — needs advancing").

Signed in? Ask the help assistant in your dashboard.

last updated 2026-06-02