Who this is for: self-serve plans (Free, Starter, Pro, Enterprise). Knowledge is included on Pro, Enterprise and the Pro trial. It is not part of the AppSumo deal; see the AppSumo documentation. Support agent: confirm whether the user is self-serve or AppSumo before giving plan-specific advice.
TL;DR: Agents → Knowledge is your workspace's own library: uploaded files, web pages, website crawls, Google Docs, plus Notion pages and Fathom meetings connected from Integrations. Clu and your agents search it and cite what they use. Admins manage it; it's free; limits are 2,000 documents and 1 GB of files per workspace. Clu in Slack doesn't use it.
Knowledge
Knowledge gives Clu and your agents your own material: product sheets, pricing pages, battlecards, sales scripts, policies, your website and notes from past meetings. When a question could be answered from that material, Clu searches it first and cites every fact as a numbered footnote linked to the document. If your Knowledge doesn't cover something, Clu says so instead of guessing.
Knowledge is about your material. Clu does not search the Cleanlist documentation.
Where to find it
Agents → Knowledge tab (
app.cleanlist.ai/agents?tab=knowledge). The Agents page tabs are Agents · Plays · Skills · Knowledge · Profile.Integrations → Knowledge section, for connecting Fathom and Notion: "Meetings and pages the whole team can search through Clu."
The Knowledge tab lists your sources in a table with Sources, Documents, Status and Last synced ("Not yet" before the first sync). Search sources filters the table by source name or address; it doesn't search inside documents. Click a source to open its details: what it covers, how often it's re-read, Who can use it, and each document with its status and size.
Who can do what:
Everyone in the workspace can view sources, documents and status.
Only workspace admins can add, rename, pause, resume, sync, change access for or remove sources ("Only organization admins can change the knowledge base."). Non-admins see Add source turned off and "An admin can add files and web pages for Clu and your agents to search."
Add a source
Click Add source on the Knowledge tab and pick one of four options.
Option | What it adds | Re-read |
Upload files | PDF, DOCX, TXT, Markdown (.md, .markdown) or HTML (.html, .htm), up to 25 MB each and 20 files at a time. All files in one upload become one source, named after the first file unless you type a Name (optional). | Never. Sync now retries files that failed. |
Add a web page | One public page, such as a pricing or product page. | Daily |
Crawl a website | Up to 200 pages of a site, found from its sitemap or from the links on the page you give. The address starts filled in with your company website from your profile. | Weekly |
Add Google Docs | Docs you pick in Google's file picker, up to 20 at a time. | Checked every 30 minutes, re-read when edited |
After you click Add, a message confirms it: "File added. Indexing it now.", "Page added. Reading it now.", "Site added. Crawling it now." or "Doc added. Reading it now."
Files: PDFs are read up to 500 pages. There is no OCR, so a scanned PDF without a text layer shows "No text found (scanned?)".
Web pages and crawls:
The address must be a public https address. Pages that need a sign-in can't be read.
Each page must load within 15 seconds, with at most 3 redirects, and be no larger than 10 MB.
Pages are read as plain HTML. Content that only appears after JavaScript runs isn't captured.
A crawl follows the site's sitemap first (from robots.txt or /sitemap.xml), otherwise links on the same site up to 3 clicks deep. It reads about one page per second, respects robots.txt and identifies itself as CleanlistAgents, so you can allow it in your robots.txt if needed.
When a crawl covers the whole site, pages that no longer exist are dropped. Any page that returns "not found" or "gone" is dropped. Unchanged pages aren't processed again.
Google Docs:
First connect Google Docs on the Integrations page (Workspace section) with your own Google account. Otherwise you'll see "Connect your Google Docs on the Integrations page first."
Cleanlist can read only the Docs you pick. Only native Google Docs can be added ("Only Google Docs can be added here.").
Docs sync through the connection of the admin who added them. If that person disconnects Google Docs, the source shows "Not connected. Reconnect it in Integrations" until they reconnect.
Connect Fathom and Notion
Go to Integrations, find the Knowledge section and click Link on Fathom or Notion. Approve access on the vendor's screen. Cleanlist creates the Knowledge source and starts the first sync right away.
Fathom | Notion | |
Card text | "Sync your Fathom meeting summaries and action items into Knowledge, so Clu can answer what was discussed with an account." | "Sync the Notion pages you choose into Knowledge, so Clu and your agents can use them." |
What syncs | Meetings the connecting member can see in Fathom (their own and ones shared with them). First sync goes back 30 days. Up to 2,000 meetings per sync. | The pages the connecting member shares with Cleanlist on Notion's consent screen. |
What's stored | Per meeting: title, date and time, duration, host, attendees, Fathom's summary, action items, topics, keywords and a link back to Fathom. No transcripts. | Each page's text. |
Sync | Every 30 minutes. Each sync re-reads from a day before the last one, so summaries written after a meeting are picked up. | Every 30 minutes: new, edited and un-shared pages. A page you stop sharing in Notion is removed from Knowledge. |
Source name | "Fathom meetings" | "Notion pages" |
Rules for both:
One connection per workspace. Anyone in the workspace can connect when nothing is connected. Only the member who connected it or a workspace admin can replace or disconnect it. Others see "Connected as ... · managed by a teammate".
Disconnect removes everything that tool synced from Knowledge and stops new syncing ("Synced meetings are removed from Knowledge and new ones stop syncing."). Fathom has no way for Cleanlist to revoke access on its side: to fully revoke, also remove Cleanlist from your Fathom settings.
If the connection breaks, the card shows Reconnect.
Want every rep's meetings? Connect Fathom with an account that can see them, since Cleanlist reads what the connecting member can see.
These are the only meeting and page connectors for Knowledge. See Google, Airtable, Notion & Fathom for what else these connections do in plays.
Sync schedule
Source | How often | Notes |
Uploaded files | Never re-read | Upload a new version to update |
Web page | Daily | A failed sync retries after an hour, then less often |
Website crawl | Weekly, up to 200 pages | Same retry rule |
Google Docs | Checked every 30 minutes | Re-read only when a Doc changed |
Notion pages | Every 30 minutes | New, edited and un-shared pages |
Fathom meetings | Every 30 minutes | Re-reads from a day before the last sync |
A source is marked Behind when it hasn't synced successfully for two of its intervals (2 days for a web page, 14 days for a crawl).
Statuses and errors
Source status: Ready, Syncing, Paused, Behind, Removing, "Some files didn't delete", or an error.
Document status: Indexed, Waiting, "No text found (scanned?)", Deleted, or an error.
Common errors and what to do:
Message | What to do |
The page needs a sign-in / The page refused access | Use a public page, or upload the content as a file |
Not a public https address | Use the page's public https address |
The page took too long to load / The page is over 10 MB | Try a lighter page or upload the content as a file |
The site's robots.txt doesn't allow it | Allow CleanlistAgents in robots.txt, or add single pages |
No pages found to read | Check the address, or add the pages one by one |
Unsupported file type / A file couldn't be read | Re-save as PDF, DOCX, TXT, Markdown or HTML |
Couldn't index it. Try Sync now | Click Sync now on the source |
Not connected. Reconnect it in Integrations | Reconnect the tool on Integrations |
Access was refused. Reconnect it in Integrations | Reconnect and approve access again |
Rate limit reached. Syncs again soon | Nothing; the next sync continues |
The page isn't shared with Cleanlist any more | Share the Notion page again |
Google won't share the Doc any more. Add it again | Add the Doc again with Add Google Docs |
The Doc is in the trash | Restore the Doc in Google Drive |
The knowledge base is full | Remove sources you don't need (limits below) |
Manage sources
Each row's menu (admins only) has:
Sync now: reads the source again right away. Only for active sources ("Resume the source first.").
Pause / Resume: a paused source is not searched and not synced, but its content is kept. Resume starts a sync.
Remove: the Remove source dialog says it "leaves search at once, and its N documents are deleted." Documents, extracted text and uploaded files are then deleted.
Removing the Fathom or Notion source in Knowledge stops its syncing until the tool is disconnected and connected again on Integrations. To stop it for a while, use Pause instead.
Who can use each source
Click a source and look at Who can use it:
Clu and every agent (the default, and where Fathom and Notion start)
Only the agents chosen: tick at least one agent ("Pick at least one agent to save.")
Changes apply from the next search.
Where the search happens | What it can see |
Clu in a regular chat (dock, Home, full screen) | Sources shared with Clu and every agent |
Clu inside an agent's workspace | Shared sources plus sources chosen for that agent |
An agent's play steps and scheduled briefs | Shared sources plus sources chosen for that agent |
In an agent's Agent tab, the Knowledge section shows Company profile (with Open to edit it in Agents → Profile), Company notes (notes Clu saved for your company), a line like "3 knowledge sources every agent can use", and a switch for each source kept to chosen agents.
How Clu and agents use it
Clu searches Knowledge before answering questions your material could cover, such as your products, pricing, positioning, policies, processes, customers and past meetings. It can also read a whole document when a passage isn't enough (you'll see "Reading a document"). Every fact it takes from Knowledge is cited as a numbered footnote with the document title linked.
Agents use Knowledge through the Knowledge tool option in a play's Agent instructions step, and in scheduled briefs when the agent can read at least one source. See Play steps.
Not used by: Clu in Slack, Clu on the Free plan, the public API and MCP.
Limits
Limit | Value |
Documents per workspace | 2,000 ("The knowledge base holds at most 2000 documents.") |
Uploaded files per workspace | 1 GB in total ("The knowledge base holds at most 1 GB of files.") |
File size | 25 MB each, 20 files per upload |
PDF length | 500 pages read |
Web page | 10 MB, 15-second load, 3 redirects |
Website crawl | 200 pages |
Google Docs per pick | 20 |
Fathom meetings per sync | 2,000 (first sync: last 30 days) |
Plans and costs
Plan | Knowledge |
Pro, Enterprise, Pro trial | Included |
Free, Starter | Not included. Upgrade in Settings → Plans & billing. |
AppSumo only | Not included ("The knowledge base isn't available on AppSumo plans."). A paid Cleanlist plan on top unlocks it. |
Costs: adding, syncing and storing sources uses no credits. When Clu searches Knowledge, it's part of that request's normal Clu cost. In a play's Agent instructions step, turning on the Knowledge tool doesn't change the step's price (3 credits per record, 5 with web search or page reading).
Privacy and data handling
Uploaded files are kept in private storage under generated IDs, never under your file names.
Each workspace's text and search index are kept separate, and every search is filtered to your workspace and to the sources that chat or agent may use.
Text is turned into a search index by an AI embedding service. Document text is never sent to Cleanlist's product analytics.
Sync errors are stored as short codes, not as your content.
Remove deletes a source's documents, extracted text and files. Disconnect on Fathom or Notion removes everything that tool synced.
Fathom and Notion access tokens are encrypted at rest.
Website crawls respect robots.txt and identify as CleanlistAgents.
See Privacy & Security for how Cleanlist handles data in general.
Related
Full documentation: docs.cleanlist.ai/features/knowledge