Give your agent long-term memory and a database

Capture sources from any channel into a per-org knowledge layer, promote them into the wiki, and read a one-call status snapshot. The primitives behind the Brain dashboard panel — and the identity-first ingest path every connector deposits through.

• -Identity-first staging (source_uri is the key, slug never auto-generated)
• -Async vectorization — searchable via search_knowledge in ~10–60s
• -Idempotent stage / promote / unstage on (source_type, slug)
• -Status snapshot is free — counts, recents, and 7-day event feed in one call

Give your AI agent long-term memory. Ingest documents, web pages, emails, and text — then search everything with natural language via vector embeddings.

• -PDF, DOCX, TXT, RTF & web page ingestion
• -Semantic search with natural language
• -Email, Slack, review & note indexing
• -Search & management tools free

Build a self-organizing wiki from raw artifacts. Ingest sources at the org or idea level — entity pages, the index, and the log update automatically. Query with citations.

• -Org-level and per-idea wiki scopes
• -Auto-updated entity pages, index, and log
• -Natural-language queries with citations
• -Query, view, and lint are all free

Save exact, user-owned files your agent gets back byte-for-byte, plus find them by meaning. The canonical-state layer of the knowledge family: Files stores what you saved, Brain derives the index, Wiki synthesizes the curation.

• -Verbatim — get_file returns the exact bytes save_file stored
• -Private & org-scoped — never a public URL, ownership checked on every read
• -Find by meaning via the same index that powers Brain
• -Folders are just path prefixes — list and find scope to a prefix

Give your AI agent a full cloud database. Import CSVs, query with SQL, insert and sync records — powered by DuckDB via MotherDuck.

• -CSV import from any URL (S3, Sheets, Drive)
• -Full SQL queries with DuckDB
• -Record insert, sync & upsert
• -Semantic search on vectorized columns

Connect a GitHub repo with a one-click app install, then read and write its docs and skill files as commits — list, get, save, append, find, and delete, all on the repo's default branch.

• -One-click GitHub App connect — no PAT to paste or rotate
• -Read & write files as real commits on the repo's default branch
• -List, get, save, append, find, create & delete files/folders
• -Org-scoped to your connected installation