Why Can't an AI Agent Find a Document You Know Is in Brain?
Work down the causes Gumloop lists when an AI agent can't find a document in Brain, starting with a sync that hasn't finished.
Check six causes in order and stop at the first that fits: the document hasn't synced, you aren't permitted to see it, the source is paused or waiting for approval, the sync failed, Brain won't read the file, or the AI agent never searched Brain. A message to the AI agent shows whether it searched at all.
You search Brain for a document you've opened in the source yourself, and the search returns nothing. Brain is available on the Pro and Enterprise plans.
In what order should you check why a document is missing?
It hasn't synced yet. External sources re-sync about every hour, and a new or edited document appears after the next sync. To update a source now, open its three-dot menu and choose Re-sync. Files you upload never sync on a schedule, so they change only when you add or remove them.
You aren't permitted to see it. With Document access set to Original source, you see only documents you could open in the original system, matched by the email you log into Gumloop with. A file shared to a different address, or only through "anyone with the link", stays hidden, and a sharing change shows after the next sync. If you can't see the source at all, you need to belong to the organization that owns it, and a team invite alone doesn't count.
The source is paused, or waiting for approval. A paused source stops syncing, but what it already indexed stays searchable, so a pause explains a missing new file and not an old one. A source in Pending approval has indexed nothing, and search returns nothing from it until someone clicks Approve and sync. Brain also pauses a source after repeated failed syncs and emails whoever added it.
The sync failed. Sync failed means a sync stopped early, and Partial sync means it finished but skipped some documents, which the source's page lists with a reason each. A failed sync removes nothing already indexed. If the source rejected its saved credentials, reconnect the source and choose Re-sync.
Brain won't read the file. File uploads take PDFs, Office files (.docx, .pptx, .xlsx), rich text, Markdown, JSON, XML, YAML and plain text, up to 25 MB each. Other types are rejected, and text inside images or scanned pages isn't read.
The AI agent never searched Brain. A source has to be indexed and then attached under Knowledge Sources before that AI agent can use it. An attachment set to Custom skips any folder or file you unchecked, and a Personal Brain source works only in your own chats.
The AI agent decides when to search, and if your credits run out, Brain searches stop working while the AI agent carries on without them. What Do I Put in an AI Agent's Knowledge Source? covers what to gather before you attach one.
Checked on 2026-11-11 against Gumloop's page on Brain and its help pages on why a document is missing, why you can't see a knowledge source, how often Brain syncs, what pending approval means, why a sync fails and how AI agents use Brain.
What message shows whether the AI agent searched Brain?
Send this in the chat of the AI agent that should have found the document. When it searches Brain, the chat shows Searching Company Brain, and when it opens a document in full, the chat shows Reading document.
Search Brain for [DOCUMENT NAME OR A PHRASE FROM IT]. I expect it in [SOURCE OR FOLDER WHERE IT LIVES]. Then answer in this order. First, say whether a Searching Company Brain step ran. Second, list the titles the search returned and the source each came from, or say "nothing returned". Third, if a title matches, open it with Read document and quote one passage from it. If nothing matches, say so and stop: don't search the web, don't answer from memory and don't guess what the document says. If no Searching Company Brain step ran, say "I can't see that" for anything about your attached sources, and change nothing in your configuration.
How do you work down the check order in Gumloop?
- 01
Read the source's status before you ask anything.
In Gumloop, open Brain from the left sidebar and click the source. The Status column and the Overview panel show whether it's Active, Paused, in Pending approval or failed, and when it last synced. Checked on 2026-11-11 against Gumloop's page on Brain and its help page on what pending approval means.
- 02
Search the source yourself before you blame the AI agent.
On the source's page, type the document's title into the Search this source box, which Brain's page describes as a preview of what's indexed. If the document comes up, it looks indexed, so go to step 03. If it doesn't, the first five causes stay open. Checked on 2026-11-10 against Gumloop's page on Brain.
- 03
Send the message and read what the chat shows.
Paste it into the chat of the AI agent. If the chat shows no Searching Company Brain step, open the tab named Agent, find Knowledge Sources, attach the source and check whether it reads All files or Custom. If the step ran and returned nothing, check that a Custom attachment includes the document's folder, then go back to the first five causes. Ask the AI agent again for the full text when it returned the title but the chat never showed Reading document. Checked on 2026-11-11 against Gumloop's help page on how AI agents use Brain, its page on Brain for the two chat labels and the tabs table on its AI agent builder page.
- 04
Re-sync the source and ask the same question again.
Open the source's three-dot menu and choose Re-sync, and send the message once the status reads Active. Re-sync is grayed out while a sync is running or a new source is waiting for approval, so check the status again. Checked on 2026-11-10 against Gumloop's help page on how often Brain syncs.
I am an ambassador for Gumloop. Code VIBE20 takes 20% off your first month.
What does a document that stays missing tell you?
Open the source's page and read the reason beside any skipped document. A document the original system refused to share stays out of search until that changes, and one skipped for a temporary reason is retried on the next sync.
If a teammate finds a document you can't, compare its sharing details for each of you.
Name Every Episode File So an AI Agent Can Find the Right One covers naming files so the right one comes back.
Frequently asked questions
Can Brain crawl a website?
The source types on Brain's page include Notion, Google Drive, Slack, GitHub, Confluence, Zendesk and file uploads, with no website crawl among them, and web-scraping platforms are hidden from the list of apps you can build a source from.
Does a paused source hide what it already indexed?
No. A paused source stops syncing, and what it already indexed stays searchable.
Why does a teammate find a document that I can't?
With Document access set to Original source, Brain matches you to a document by the email you log into Gumloop with. Compare the sharing details for each of you.