Your Learning Management System (LMS) Admin wants Aura to answer your people's questions from documents and pages you already keep in SharePoint, Google Drive, OneDrive, Dropbox, Box, Confluence, or Notion. This article covers how the Connected Knowledge connection works, what it can reach, and what you control, so you can review and approve it.
What You're Approving
Read access to the content in scope, through one account you own. You create a dedicated integration account and share only the folders, spaces, or pages your LMS Admin plans to connect. That account signs in once through standard Open Authorization (OAuth). From then on, Absorb reads from that content on a schedule and never writes back.
The key points are summarized in the callout below.
- Reads, never writes: Absorb doesn't create, edit, or delete anything in your repository.
- You set the outer edge: The connection reaches only what the integration account can reach.
- Closed by default: Nobody gets answers from synced content until an LMS Admin adds a rule for them.
- Never used for training: Your documents answer your people's questions and never train artificial intelligence (AI) models.
- One partner in the middle: Merge, Absorb's integration partner, brokers each connection. It keeps file details, plus page copies for Confluence and Notion.
- Off in one step: Revoke the account's access, or delete the connection in Absorb.
How the Pieces Fit Together
Content moves one way, from your repository into Absorb. Absorb never writes anything back. The following table outlines the four components involved:
| Component | Role | What It Does |
| Your repository | Source | SharePoint, Google Drive, OneDrive, Dropbox, Box, Confluence, or Notion. You pick the repository and the account that signs in. Your existing sharing settings decide what that account can see. |
| Merge | Connector | Holds the OAuth connection and brokers each download. Merge is Absorb's integration partner for repository connections. Merge authenticates with your repository and keeps the details needed to sync, like names, folders, and dates. For Confluence and Notion, it also keeps a copy of page content. |
| Absorb | Platform | Syncs, extracts the text, and applies your Availability Rules. Absorb receives each selected file or page, extracts the supported text, and stores it in Absorb-managed systems. Absorb keeps each customer's data separate from every other customer's. |
| Aura | Answer | Answers your people and names the source document. Aura finds the passages that answer a question, then writes a reply that names the document it came from. |
Two Kinds of Connection
File storage and knowledge bases answer your people the same way. They differ in what Merge keeps along the way.
| Connection Factor | File Storage | Knowledge Base |
| Repositories | SharePoint, OneDrive, Google Drive, Box, Dropbox. | Confluence, Notion. |
| Who it serves | Your people, through Aura answers, plus course authors in Aura Create. | Your people, through Aura answers, plus course authors in Aura Create. |
| How it connects | Through Merge. | Through Merge. |
| What Merge stores | File, folder, and drive details only, never file content. | Pages, their details, and a copy of page content. |
| What you pick | Sites, drives, folders, and files. | Spaces and pages in Confluence. Pages and databases in Notion. |
| How content reaches Absorb | Files stream through Merge in real time. | Absorb pulls each page from Merge's stored copy. |
What Happens When Someone Asks
Absorb checks access on every question, before Aura reads a single passage. Here's the path a question takes.
- Someone asks Aura a question. They can ask wherever they use Aura. The same checks run behind every place they ask.
- Absorb checks who they are. Absorb matches the person's Absorb Groups and Departments, the ones set up in your LMS, against the Availability Rules on each connection. These aren't SharePoint, Google, or other repository groups. Absorb reads the rules at the moment of the question, so a change in Absorb counts right away.
- Absorb searches only what they're allowed. The search covers just the files, folders, spaces, and pages those rules share with that person. No matching rule means no results from that connection.
- Aura writes the answer. Absorb sends the question and the few passages that match to the AI provider. Absorb never sends your repository as a whole, and never sends any file or page the person's rules leave out.
- The answer names its source. Your people see which document or page the answer came from, so they can go to the original for full context.
Two boundaries decide what anyone sees. What you share with the integration account sets the outer edge. Inside it, Availability Rules built on Absorb Groups and Departments decide who gets answers from which content. Aura doesn't read each person's own permissions in your repository, so keep anything you don't want to share off the integration account.
AI Providers and Guardrails
Aura runs on OpenAI and Google Gemini. Your content only answers questions and never trains AI models. The AI Features Addendum covers the details.
Aura screens questions and answers. A flagged request gets a safe refusal instead of an answer.
Choose The Account That Signs In
The connection sees exactly what the signing-in account sees, and nothing more. That makes the account your main control.
User Account or Admin Account
For file storage, your LMS Admin selects one in Absorb before signing in, so tell them which one you set up.
| Account Type | Reach |
| User account | Narrowest reach. Reaches the account's own files plus anything shared with it. Pair it with a dedicated integration account and share only the folders in scope. |
| Admin account | Reaches most or all of the workspace or tenant, and reads Users and Groups. The account must hold Admin rights, or the connection fails or comes back missing data. |
Every connection requests read-only scopes except Box. For Confluence and Notion, the connection sees what the signing-in account can open in that site or workspace. The Scopes by Repository section lists the scopes and explains the Box exception.
Use a Dedicated Integration Account
A non-personal account, such as aura-connect@yourcompany.com, keeps the connection steady and easy to audit.
| Connection Factor | A Named Employee's Login | A Dedicated Integration Account |
| Continuity | Sync stops when the person changes roles or leaves. | Unaffected by staffing changes. |
| Reach | Everything that person can open, personal drives included. | Only the folders, spaces, or pages you grant it. |
| Audit trail | Repository logs show every read as that employee. | Reads show under the integration account. |
| Relinking | Needs that specific person. | Any Admin with the credentials. |
Before the Sign-In
Complete the following before the sign-in:
- Allow pop-ups for your Absorb Portal domain. Sign-in opens in a second window.
- Make sure multi-factor authentication (MFA) and conditional access allow the account one interactive sign-in.
- Check whether your repository charges a seat for a service account.
- If your Atlassian organization or Notion workspace limits third-party apps, approve the Merge app before sign-in.
- Name an owner for the account and store its credentials in your password manager.
- Expect Merge's name on the consent screen, with links to Merge's privacy policy and terms.
- Whoever completes the sign-in should use a private window, so the browser doesn't reuse their own session.
Scopes by Repository
The following table lists the scopes the connection requests for each repository and account type. The consent screen shows the final list for your tenant or workspace.
| Repository | User Account | Admin Account |
| Google Drive | drive.readonly | drive.readonly admin.directory.group.readonly |
| SharePoint | Files.Read.All Sites.Read.All User.Read User.ReadBasic.All |
Files.Read.All · Sites.Read.All Group.Read.All · GroupMember.Read.All User.Read.All RoleManagement.Read.All RoleManagementPolicy.Read.Directory |
| OneDrive | Files.Read.All Group.Read.All · GroupMember.Read.All User.Read · User.ReadBasic.All |
Not offered. Use the User account option. |
| Dropbox | account_info.read files.metadata.read files.content.read sharing.read |
The four user scopes, plus: team_info.read · team_data.member team_data.governance.read team_data.content.read files.team_metadata.read members.read · groups.read · events.read |
| Box | root_readonly · root_readwrite manage_managed_users manage_groups · manage_webhook |
Same as the User account. |
| Confluence | Read-only access to spaces, pages, and their details, as Atlassian's consent screen lists them. | |
| Notion | Read content, limited to the pages you pick on Notion's consent screen and the pages under them. | |
What the Scopes Are For
The scopes cover reading file and folder details and content, listing drives and sites for the picker, and reading basic User and Group details. Admin scopes add User and Group lookups and a check that the account holds Admin rights.
The following notes apply to specific repositories:
- Box is the exception: Box has no read-only scope that lets an app download files, so its consent screen lists read and write scopes. Absorb only reads. Limit the account's Box access to the folders in scope.
- Microsoft 365: These are delegated Microsoft Graph permissions, so the connection acts as the signed-in account. Group.Read.All, GroupMember.Read.All, and User.Read.All need Admin consent in Entra ID.
- Google Workspace: The group scope needs a Workspace Admin to sign in. If your Workspace restricts third-party apps, allow the Merge app in application programming interface (API) controls before sign-in.
- Dropbox: Dropbox requires every app to name its scopes, so the list on your consent screen may differ slightly from this one. Proceed with the list shown on your screen.
Where Your Data Goes and Stays
Absorb keeps only what it needs to find answers in the content you selected. The following table outlines each type of data, where it lives, and why.
| Data | Where It Lives | Why It's There |
| File, folder, space, and page details | Merge and Absorb. | Names, locations, dates, and types identify and sync the items you selected. |
| File content | Streams through Merge, which doesn't keep a copy. | Absorb receives the file and extracts the supported text. |
| Extracted text | Absorb-managed systems. | Absorb splits the text into searchable passages so Aura can find and cite answers. |
| Confluence and Notion page content | Merge and Absorb. | Merge keeps a copy so Absorb can sync it, and removing the connection deletes that copy. |
| OAuth credentials | Merge | Keep the connection signed in. Encryption includes an added field-level layer. |
How Syncing Works
The following points describe how syncing works:
- Absorb checks each connection for changes every hour.
- Absorb processes again only the new and changed files and pages.
- Deleted files and folders drop out on the next sync.
- If a run fails, the next hourly run picks up the same changes.
- Changes reach Aura in about an hour, longer for large updates.
What Syncs
The following points describe which content syncs:
- PDF, Word (.docx), Excel (.xlsx), PowerPoint (.pptx), and Google Docs, Sheets, and Slides.
- Absorb skips older .doc, .xls, and .ppt files, and scanned PDFs with no text layer.
- For Confluence and Notion pages, Absorb syncs the text in the page body.
- A selected folder, space, or parent page includes what's inside it, including items added later.
The connection screens link the data processing agreement next to Merge's privacy policy and terms. For the AI Features Addendum, the subprocessor list, data residency, and retention schedules for your contract, ask your Absorb Account Manager.
Ways to Limit or Switch It Off
You control your repository, and your LMS Admin controls Absorb. Controls are available to both you and your LMS Admin to limit or disable a connection.
In Your Repository
The following table outlines the controls available to you.
| Control | What It Does |
| Grant or remove access | Change which folders, spaces, or pages the integration account can open. This is the lasting boundary. |
| Revoke the app's access | Remove the OAuth grant to stop syncing right away. |
| Disable the account | Ends the connection until someone relinks with a working account. |
| Review audit logs | Every read shows under the integration account in your repository's own logs. |
In Absorb
The following table outlines the controls available to your LMS Admin.
| Control | What It Does |
| Availability Rules | Decide which Absorb Groups and Departments get answers from which content. Rules only add access. |
| Edit connection | Change which folders, files, spaces, or pages sync. Removed items drop out of every rule. |
| Relink integration | Sign in again after a password or credential change. |
| Delete connection | Removes the connection and every rule on it. Answers can no longer be created from files within this connection. |
Connected Knowledge in Aura Create
Course Authors in Aura Create can draw on everything a connection syncs, whatever the Availability Rules say. Choose what to sync with both audiences in mind.
Disable Aura Features
Each Aura product can be turned off in Absorb on its own, without touching the connection. In the Admin Experience, access the Aura Configuration page through the Client Settings to disable Aura features.
Quick Fixes
The following table outlines common problems and how to resolve them.
| If You See | Try This |
| The sign-in window never opens | Allow pop-ups for your Absorb Portal domain, then click Open window again. |
| Consent is blocked | Have an Entra ID, Workspace, Atlassian, or Notion admin approve the Merge app, then sign in again. |
| A folder or page is missing from the picker | Share it with the integration account (in Notion, share the page with the connection), then reopen the picker. |
| The Last synced date stops moving | Check the account is active and its grant is in place, then click Relink integration. |
Your Part in the Rollout
You handle the first three steps. Your LMS Admin takes it from there. For more information about setting up Availability Rules and managing connections in Absorb, see the Aura Learner Agent: Connected Knowledge article.
- Approve read access: Review this article and sign off on the content in scope.
- Set up the account: Create the integration account and share only that content with it.
- Clear the sign-in: Ready Admin consent, MFA, and pop-ups so the sign-in goes through.
- LMS Admin connects: Signs in with the account, picks what to sync, and sets up rules.
A Summary for Your Change Record
Fill in the fields in angle brackets and keep this with your change ticket or security review.
| Change | Read access for Absorb Aura Connected Knowledge |
| Repository | <SharePoint | Google Drive | OneDrive | Dropbox | Box | Confluence | Notion> |
| Content in scope | <list the exact sites, drives, folders, spaces, or pages> |
| Account | Dedicated integration account, suggested aura-connect@yourcompany.com. Owner: <name or team> |
| Account type in Absorb | <User account | Admin account> for file storage, read-only scopes (Box lists read and write) |
| Sign-in | One interactive OAuth consent. MFA or conditional access exception if needed. Allow pop-ups for <your Absorb portal domain> |
| Third party | Merge brokers the connection and appears on the consent screen. For Confluence and Notion, Merge keeps a copy of page content. |
| Data handling | Absorb extracts and stores text from selected files and pages so Aura can answer and cite it. Absorb writes nothing back, and content never trains AI models. |
| Who sees it | No one until an LMS Admin adds Availability Rules. |
| Removal | Revoke the account's access, or delete the connection in Absorb. |
Your Sign-Off Checklist
Use the following checklist for your sign-off:
- ☐ You've approved read access to the named repository.
- ☐ The integration account exists, has a license if needed, and has a named owner.
- ☐ It can open exactly the content in scope, and nothing more.
- ☐ MFA and conditional access allow one interactive sign-in.
- ☐ Admin consent or app approval is ready where your repository needs it.
- ☐ You've reviewed Merge's terms, and you expect its name on the consent screen.
- ☐ You're comfortable that Course Authors in Aura Create can use all synced content.
- ☐ You've set a date to review scope and access again.
Answers for Your Review
Here are short answers to the questions that come up most in security reviews.
Can Aura Change or Delete Our Files?
No. Absorb only reads. In Box, the consent screen lists write scopes because Box has no read-only download scope, but Absorb doesn't use them.
Does Aura Follow Each Person's SharePoint Permissions?
No. Availability Rules built on Absorb Groups and Departments decide who gets answers, inside what the integration account can open.
Why Does the Consent Screen Say Merge?
Merge is Absorb's integration partner for repository connections. It holds the OAuth connection and brokers downloads, and for Confluence and Notion it keeps a copy of page content. That screen links its privacy policy and terms.
Is Our Repository Sent to an AI Model?
No. For each question, Absorb sends the question and the few passages that match it. Only content the User's rules allow can be among them.
Can We Limit It to One Site, Folder, or Space?
Yes. Share only that content with the integration account. Your LMS Admin can narrow it further when picking what to sync.
How Current Are the Answers?
Absorb checks every hour. A new or edited file or page usually reaches Aura within about an hour, longer for large updates. Deleted files drop out on the next sync.
What if the Account's Password Changes?
Syncing stops and the Last synced date stops moving. Your LMS Admin clicks Relink integration and signs in again with the same account.
Do We Need Admin Consent?
Often, yes. Several Microsoft Graph scopes need an Entra ID admin, and Google's group scope needs a Workspace admin. Plan that approval before the sign-in.
Who Else Can See Synced Content?
Course Authors in Aura Create can draw on everything synced to help build courses. Availability Rules apply to answers for your people, not to authors.
Where Is It Stored, and for How Long?
Synced text lives in Absorb-managed systems, separate for each customer. Removing the connection deletes Merge's copy of Confluence and Notion pages. For data residency and retention schedules on your contract, ask your Absorb Account Manager.
How Do We Shut It Off?
Revoke the app's access in your repository, or have your LMS Admin delete the connection in Absorb. Deleting removes the connection and every rule on it.
Is Our Content Used to Train AI Models?
No. Your documents only answer your people's questions. They never train AI models. The AI Features Addendum sets out how AI providers handle your content.
Questions along the way? Your Absorb Customer Success Manager can walk you through the connection.
Comments
Article is closed for comments.