Skip to content

Knowledge

Accessible to: Workspace Owners, Admins, Trainers

The Knowledge section is your workspace's own library of documents. You upload the policies, handbooks, procedures and decks your training is actually based on, and Aimee keeps them and makes them searchable.

This is different from the reference files you can attach in the Generate with AI wizard. Those are read once, summarised, and thrown away. Documents you add here are kept, indexed, and searchable at any time.

Turn your Knowledge into a course

Once your material is here, Aimee can build a course from it - grounded in your documents, with a citation on every slide, and kept current as the source changes. See Building a Course from Knowledge.

Turned off until an owner turns it on

Knowledge does not appear in the sidebar by default. A workspace owner or admin switches it on in Workspace Settings → Features using the Knowledge Base switch. Until then, nobody in the workspace sees the Knowledge entry.

You will find Knowledge in the left sidebar, just below Content Library.

The page is organised into tabs: Sources lists every document in the knowledge base, Search is where you look things up, and - where an owner has enabled connectors - Connectors manages automatic syncing from SharePoint, Confluence and Google Drive.

Adding documents

Select Add documents at the top right and choose your files. You can add up to five files at a time, totalling no more than 25 MB.

Supported file types:

  • PDF (.pdf)
  • Word documents (.docx)
  • PowerPoint decks (.pptx)
  • Plain text (.txt)
  • Markdown (.md)

Adding a document does not block you. Each file appears in the list straight away with a Queued label, moves to Indexing while Aimee reads and indexes it, and finishes at Indexed. Large files take longer; the page updates itself while anything is still in progress, so you can leave it open and carry on.

Uploading the same file twice is free

If you add a document that is already in your knowledge base and has not changed, Aimee recognises it and does nothing - you are told it is already there. If the document has changed, the new version replaces the old one for searching, and search results stop showing the outdated text.

What the list tells you

Each row shows:

  • Source - the document's title and original filename.
  • Status - Queued, Indexing, Indexed, or Failed.
  • Documents and Chunks - how many pieces the document was broken into for searching. A long policy produces more chunks than a one-page memo.
  • Size - how much of your workspace's storage allowance it uses.
  • Added - when it was uploaded.

The four tiles at the top of the page summarise the same figures across your whole knowledge base.

If a document fails

A document can fail if Aimee cannot read any text from it - a PDF that is purely scanned images is the most common cause. Hover the Failed label to see why. Select the refresh icon on that row to try again; this is worth doing if the problem was temporary.

Connecting SharePoint

Uploading files by hand works well for a handful of documents. When the material your training relies on lives in SharePoint and changes regularly, a connector keeps the knowledge base up to date automatically: it syncs the document libraries you choose once a day, picks up new and edited files, and removes anything deleted at the source.

Two switches, and who can connect

Connectors have their own switch in Workspace Settings → Features - Knowledge connectors - which only becomes available once the Knowledge Base itself is on. Only workspace owners and admins can connect, configure or disconnect a connector; trainers can see connectors and their sync history but not change them.

Everything synced is visible to the whole workspace

A connector does not copy SharePoint's own permissions. Every document it brings in can be found by anyone in this workspace who can use the knowledge base - so only connect libraries and folders that everyone here may read.

Your workspace's own OAuth apps

This step requires IT or developer access

Setting up an OAuth app involves creating an application registration with your cloud provider (Microsoft, Atlassian, or Google) and generating client credentials. This is a one-time technical task - if you're not comfortable in your cloud provider's admin console, ask your IT team or a developer to do this part. Once the app is registered and its credentials are saved, you won't need to repeat it.

Connectors sign in through OAuth apps your workspace owns. Before a Connect button appears for SharePoint, Confluence Cloud or Google Drive, an owner or admin registers an app with that provider and saves its credentials under Workspace Settings → Integrations → Connector OAuth Apps:

  • SharePoint: an Azure App Registration with delegated read permissions (Files.Read.All, Sites.Read.All).
  • Confluence Cloud: an Atlassian OAuth 2.0 (3LO) app with the classic read scopes.
  • Google Drive: a Google Cloud OAuth client with the read-only Drive scope. Google marks this scope restricted: while your app is unverified it works in testing mode for the accounts you list as test users, which is enough for most teams.

Each provider's app needs the redirect URI shown in that settings panel. The client secret is stored encrypted and never displayed again; replacing or removing an app is done in the same panel. Self-hosted Confluence (Data Center) needs no app - it connects with a Personal Access Token directly.

Connecting

On the Connectors tab, select Connect SharePoint. You will be sent to Microsoft to sign in and approve read access, then brought straight back.

Next, choose what to sync: browse your sites, pick a document library, and optionally narrow down to a single folder. You can add several locations to one connector.

Before anything is ingested, select Preview. It shows how many files the selection contains, their combined size, a breakdown by file type, and what would be skipped and why. Nothing is synced and nothing is spent until you select Start syncing - so a much larger library than you expected is something you see on screen, not something you discover afterwards.

What syncs and what is skipped

Connectors sync the same file types you can upload by hand - PDF, Word, PowerPoint, plain text and Markdown. Anything else (spreadsheets, images, site pages) is skipped, as is any file over 25 MB. Skipped files are always counted and shown, in the preview and in every sync's results, so nothing disappears silently.

A file that is byte-for-byte identical to something already in your knowledge base is recognised and not indexed twice.

Day to day

Each connector syncs once a day on its own, and owners and admins can run Sync now at any time. The connector row shows when it last synced and what happened: how many documents were added, updated, removed and skipped.

Documents a connector brings in appear on the Sources tab alongside your uploads, labelled with the connector's name, and behave exactly the same in search.

When a file changes in SharePoint, its indexed copy is replaced at the next sync and search stops showing the outdated text. When a file is deleted in SharePoint - or moved out of the synced folder - it leaves the knowledge base at the next sync too.

If a connector stops working

If the Microsoft sign-in behind a connector expires or is revoked, the connector shows Reconnect needed and pauses, and the person who created it receives an email with a reconnect link - this is not something that fixes itself, so it is flagged loudly. Documents already synced stay available; they just stop updating until someone reconnects.

A sync that fails for a temporary reason shows Sync failed with the error; the next daily sync usually clears it.

Disconnecting

Select the disconnect icon on the connector's row. You will be asked what to do with the documents it synced: keep them in the knowledge base as static copies that no longer update, or remove them along with the connector. Removed documents follow the same trash-retention window as any other deletion.

Connecting Confluence

If your team's knowledge lives in a Confluence wiki, a Confluence connector brings whole spaces into the knowledge base: every page and blog post is converted to searchable text, and the files attached to them sync just like uploads. The same rules as SharePoint apply - the connector is read-only, owners and admins manage it, and everything it brings in is visible to the whole workspace.

Confluence Cloud or self-hosted

  • Confluence Cloud (your wiki lives at something.atlassian.net): once your workspace's Atlassian app is saved in Settings → Integrations, select Connect Confluence, sign in with Atlassian and approve read access, and you are brought straight back.
  • Self-hosted Confluence (Data Center): select Connect Confluence Data Center and enter your wiki's web address and a Personal Access Token. Create the token in Confluence under your profile's Personal Access Tokens page - read permissions are enough. The address must start with https:// and be reachable from the internet; a wiki that is only available inside your office network or VPN cannot be synced. The token is stored encrypted and never shown again.

Choosing spaces

After connecting, pick the spaces to sync. As with SharePoint, Preview shows what the selection contains - how many pages, blog posts and attachments, and what would be skipped - before anything is ingested or any indexing cost is incurred.

What syncs

Pages and blog posts arrive as text documents named after their titles; tables and code blocks survive, while decorative elements are reduced to their text. Attachments follow the same rules as SharePoint files - PDF, Word, PowerPoint, plain text and Markdown up to 25 MB; everything else is skipped and counted.

One difference worth knowing

Confluence tells us about new and edited content immediately, but not about deletions. A page deleted (or moved out of a synced space) leaves the knowledge base within a week on the daily schedule. If you need it gone sooner, run Sync now - a manual sync always does the full check.

If the connection stops working - the Atlassian sign-in is revoked, or a Personal Access Token is revoked or expires - the connector shows Reconnect needed and the person who created it gets an email, exactly as with SharePoint. For a self-hosted wiki, reconnecting means entering a fresh token.

Connecting Google Drive

If your material lives in Google Drive - including documents written in Google Docs - a Drive connector keeps chosen folders in sync with the knowledge base. The same rules as the other connectors apply: it is read-only, owners and admins manage it, and everything it brings in is visible to the whole workspace.

Connecting a Google account

Once your workspace's Google OAuth client is saved in Settings → Integrations, select Connect Google Drive on the Connectors tab, sign in with Google and approve read access. You are brought straight back to pick folders.

Choosing folders

The picker shows My Drive and any shared drives you can see; browse into either and add the folders to sync. Subfolders are always included. As with the other connectors, Preview shows what the selection contains - and what would be skipped - before anything is ingested or any indexing cost is incurred.

What syncs from Drive

  • Google Docs arrive as text documents named after their titles.
  • Google Slides arrive as presentations and are indexed like PowerPoint files.
  • Regular files in the folder - PDF, Word, PowerPoint, plain text and Markdown up to 25 MB - sync exactly like uploads.
  • Google Sheets and shortcuts are skipped and show up in the skipped count, so nothing disappears silently.

Moved folders take a little longer

Google tells us about new, edited and deleted files immediately, with one exception: when a whole folder is dragged out of a synced folder, the files inside it can linger. They leave the knowledge base within a week on the daily schedule - or immediately if you run Sync now, which always does the full check.

If the Google sign-in is revoked, the connector shows Reconnect needed and the person who created it gets an email, exactly as with the other connectors.

Searching

The search box is the quickest way to check that a document is properly indexed, and it is useful in its own right for finding a passage you half remember.

Search works two ways at once:

  • It understands meaning, so you can describe what you are looking for in your own words even if the document phrases it differently.
  • It also matches exact terms, so searching for a policy number, a product name or an acronym finds the paragraph containing it.

Each result shows the passage itself, which document it came from, the heading or slide it sits under, and the page number where there is one - so you can go straight to the source and check it.

If a search returns nothing, either no document covering that subject has been added yet, or the document you expect is still indexing.

Removing a document

Select the bin icon on a row and confirm. The document stops appearing in search results immediately.

It is not destroyed straight away: it stays recoverable for the same trash-retention period your workspace uses for deleted courses, set in Workspace Settings. After that window, the document and everything indexed from it are deleted permanently.

Course health

A course built from your knowledge base keeps a link back to the passages behind each slide. Those passages can move underneath it: a document is replaced by a newer version, a source is removed, or a connector stops syncing a page. When that happens the published course is still teaching what the old passage said, and nothing on the course itself shows it.

Course health is the list of courses that has happened to. Owners, admins and trainers get a Course health entry in the sidebar once the feature is switched on in Workspace Settings > Features. Learners never see it.

Aimee checks once a day. Each course on the list shows:

  • how many slides are affected;
  • which knowledge base documents changed underneath them, and in what way - replaced by a newer version, source removed from the knowledge base, or the cited passage no longer indexed at all;
  • when the change was first noticed.

Select a course title to open it and review the affected slides against their citations in the slide editor.

If a change does not actually affect what the course teaches - a formatting-only revision, say - select Dismiss on that document. It leaves the list and stays gone, unless that same passage changes again later, when it comes back.

Only published courses are checked. A draft is not live, so a stale citation on one is not yet a problem.

The list is empty when every published course still cites live documents, which is the normal state.

Storage and cost

Documents in your knowledge base count towards your workspace's total storage allowance, the same as course media does. If you reach the limit, further uploads are refused until you remove something or your allowance is raised.

Indexing a document also uses a small amount of your AI allowance, which is why the feature is off until an owner enables it, and why an unchanged re-upload deliberately costs nothing.