October 13, 2026
Release version: 562
This release brings Deep Research to the skill picker for Glean Universal Key customers, adds the Microsoft Fabric IQ MCP template, and makes GPT-6 series models available through custom providers. It also adds Loopio connector setup, agent sharing and API token controls, clearer Azure Files search, and companion and realtime voice updates, alongside admin, security, and reliability improvements.
Major Launches
Connect Glean to Microsoft Fabric IQ
Admins can use the Microsoft Fabric IQ MCP template to discover and query Power BI reports and semantic models through Microsoft's Fabric IQ endpoint. The template gives teams a supported starting point for connecting Glean to Fabric IQ and using the resulting tools through an MCP server, subject to the required Microsoft and Glean permissions.
How to access: In the Admin console, open the MCP server setup experience and choose the Microsoft Fabric IQ template. Complete the Microsoft authorization and connection steps.
Start Deep Research from the skill picker
Deep Research is a skill in chat for Glean Universal Key customers. Open the skill picker and select Deep Research, or type /deep-research and your question. The report is still a cited write-up from company sources and the web, and it still runs in the background until Glean notifies you.
On the Glean Universal Key, Adaptive, Fast, and Thinking are no longer choices. Glean routes each question, and Deep Research uses standard Assistant terms. Use the Deep Research skill when you want the long report.
If Deep research is still in the mode picker, your organization has not moved yet. Customer Key organizations keep Deep Research as a chat mode with the existing monthly quota until migration.
A chat that used the previous Deep Research agent shows a banner. Start new chat fills in /deep-research.
How to access: In Chat, select Deep Research from the skill picker, or type /deep-research and your question. An admin has to turn Deep Research on.
AI & Model Updates
Choose GPT-6 series models with custom providers
GPT-6 series models are not automatically available to every customer. Organizations using supported, configured custom model providers can now select GPT-6 series models where those models are available. This expands model choice for teams that manage their own provider connections while preserving the provider, permission, and configuration requirements already in place for custom models.
How to access: In the Admin console, open your custom provider configuration and select a supported model from the models available to that provider.
Admin Features
Retire the Deep Research monthly quota on the Glean Universal Key
The per-user monthly quota, which defaults to 10 runs and has a maximum of 50, is retired for Glean Universal Key organizations. Saved quota values do not block skill runs. Glean routes each Deep Research question, and a run uses standard Assistant terms. On Enterprise Flex, standard models follow Assistant terms, and premium models consume FlexCredits. You can still turn Deep Research on or off, including for a subset of users.
Customer Key organizations keep Deep Research as a chat mode with the existing monthly quota until migration. On the skill, Deep Research uses the model selected for that chat. On the Glean Universal Key, that is the model Glean routes to.
How to access: Go to Admin console → Platform → Assistant → Settings and open Deep Research.
Enhancements
Configure Loopio as a knowledge connector
Admins can now configure Loopio in the admin console so Glean can connect to Loopio content as a knowledge connector. This gives your team another way to bring relevant Loopio knowledge into Glean search and AI experiences while keeping setup in the same place as other connectors. An admin may need to enable the connector, and availability may depend on rollout.
How to access: In the Admin console, open the connector setup area and choose Loopio. Follow the connection and configuration steps shown there.
A Halloween surprise on the homepage
On October 30 and 31, click the greeting on the Glean homepage to reveal a seasonal swarm of bats. This lighthearted experience is available for a limited time and does not change search, content, or admin settings.
How to access: Open the Glean homepage on October 30 or 31, then click the greeting.
Fix private GitHub skill imports
Private GitHub skill imports now request the correct connector authorization, use configured GitHub connector credentials when available, and retry after sign-in when access requires it. This fixes an import issue for skills that depend on private repositories and makes the recovery path clearer. Connector permissions still determine which repositories and content Glean can access.
How to access: In the Skills experience, start importing a skill from a private GitHub repository. Follow the GitHub connection or sign-in prompt if it appears.
Keep embedded agent chats focused
The Web SDK's restrictToApplication option now keeps an embedded chat focused on a single agent. Users see that agent's chat history and a New chat button without the general navigation or chat header, making embedded experiences more focused and consistent.
How to access: Configure restrictToApplication in your Web SDK chat options for the embedded agent experience.
Get email updates when usage-limit requests are approved
When an agent or user usage-limit increase request is approved, the relevant requester receives an email with the updated limit and a direct link to view usage. Agent owners and editors are also notified when an agent's limit changes. Workspace admins, super admins, and billing moderators can also receive a daily email summary when spend-limit increase requests are pending. These updates make it easier to track limits without checking the admin console manually.
How to access: No action is needed to receive an approval email. Admins continue to review usage-limit requests through the existing usage-limit workflow.
Review downvoted Slack responses by email
Customers who have chat feedback emails enabled can now receive email notifications when someone downvotes a Glean response in Slack. The email includes the submitted feedback, the question, the downvoted response, and a link to the Slack message when available. In channels and group DMs, the email includes the downvoted exchange; one-to-one DMs may include earlier turns when available.
How to access: An admin must enable chat feedback emails. After that, eligible downvotes in Slack are included in the configured feedback notifications.
Search Azure Files with richer filters and metadata
Azure Files search now supports more consistent file-type handling, created-date filtering, share and connector-instance filters, and clearer result metadata. Results can show creation and modification dates, owner, path, department, file share, storage account, and custom metadata tags. Links also follow the configured file or HTTPS URL format, while document permissions remain enforced.
How to access: Search content from an Azure Files connector, then use the available filters and result details. Use in: for a share or appinstance:"" for a uniquely named connector instance.
Use Glean companion in supported browsers
Glean companion is a browser-extension widget that works on highlighted text in supported browsers. You can use it to explain terms, find related documents and experts, improve writing, summarize content, or translate without leaving the page. Glean companion is turned off in Brave and may require additional browser permissions on some webpages.
How to access: Install the Glean browser extension, open a supported webpage, and highlight text. Select the Glean icon that appears next to the highlighted text.
Make realtime voice easier to discover
A one-time in-product tip now appears on the voice button for eligible Glean key customers. It introduces the more natural, low-latency realtime voice experience so users can recognize when spoken, back-and-forth interaction is available. Only eligible organizations with the feature enabled can use realtime voice.
How to access: Open Glean and select the voice button when realtime voice is available for your workspace.
Create API tokens for agents from the share dialog
Agent owners and editors can create and manage an API token directly from an agent's share dialog. An agent API token provides programmatic access to that agent. The API area now appears first in publishing options and shows the token status with Create token and Delete actions. Admins can turn off editor token creation so only admins and agent moderators can create tokens.
How to access: Open an agent's share dialog, select the API publishing option, and use Create token or Delete when permitted by your admin settings.
Find sharing controls faster in the agent library
Agent library cards and the agent details view now place Share next to Duplicate, while the View usage icon is removed from those headers. The updated layout makes the most common publishing and collaboration action easier to find without changing the agent's permissions.
How to access: Open the agent library, choose an agent, and use Share from the card or agent details view.
Agent access granted notifications on Slack
When you receive access to an agent through direct sharing or an approved access request, Glean can now send a Slack DM in addition to the existing email notification. When sharing a chat about an agent, the share dialog can also grant the recipient Viewer access to that agent directly. These updates make access changes easier to notice and complete in one workflow.
How to access: Enable Slack notifications to receive access DMs. When sharing an agent chat, use the Viewer access option in the share dialog when it is available.
See shared work in trending discovery
Discovery experiences that explicitly request shared work can now include shared-work signals in trending results. Document access permissions continue to apply, so users only see content they are allowed to view. This helps relevant shared work surface in discovery experiences without changing the underlying permissions on that content.
How to access: Use a discovery experience that requests shared work. No additional setup is required for users.
Connect Gemini to Glean through MCP
Glean now provides a dedicated MCP server for Gemini with core Glean tools, including search, document reading, code search, and people search. Admins can also copy a Gemini tool specification from the MCP server page and paste it into Google's Gemini Agent Registry. This gives teams a more direct setup path for connecting Gemini to authorized Glean content.
How to access: In the Admin console, open the MCP server page for Gemini. Use the provided setup details, then copy the tool specification into the Gemini Agent Registry when configuring the connection.
Clarify Box API allocation settings
The Box connector setup field formerly labeled Monthly QPS allocation is now labeled Monthly API call allocation. The new label makes clear that the value represents a monthly call count, not a per-second rate, so admins can configure the connector with less ambiguity.
How to access: Open the Box connector configuration in the Admin console and use Monthly API call allocation.
Keep topic memories aligned with agent permissions
Topic memories for identity-backed agents now update to match the agent's current viewer, editor, and general-access permissions after publication and later access changes. Newly created topic memories are also updated asynchronously. This helps keep memory access aligned with the permissions of the agent that uses it.
How to access: No user action is needed. Agent and memory permissions are synchronized as access changes are published.
Strengthen GCP Node Agent deployment identities
GCP deployments now provision a dedicated service account and Cloud SQL IAM user for Node Agent. This creates a narrowly scoped deployment identity with access to the frontend Cloud SQL instance, deployment configuration, key management, and observability services. The identity is provisioned on every deployment; Node Agent pods use it when the Node Agent capability is enabled.
How to access: No end-user action is needed. GCP administrators can review the deployment configuration and Node Agent setting in their existing deployment workflow.
Improve upgrade reliability for private-beta tools
Deployment upgrades now preserve the capitalization of existing tool names. This prevents affected private-beta tools from becoming unavailable when Code Writer tools are added during an upgrade.
How to access: No action is needed. The protection applies during supported deployment upgrades.
Attribute MCP tool calls more accurately in Insights
MCP Insights now credits calls made through run_tool to the tool that actually ran, such as a Jira or Slack tool under its MCP server name, instead of attributing the call to run_tool itself. Each dispatched tool call is counted once, historical run_tool entries are no longer shown, and usage events include each tool dispatched through run_tool.
How to access: Open MCP Insights in the Admin console. No additional setup is required.
Capture more agent context in customer event logs
Customer event log entries now include an agent's WorkflowId and an agent schedule's SubscriptionId when an action is taken on a specific agent or schedule. These identifiers help administrators and support teams trace activity to a specific agent run or scheduled trigger.
How to access: Review customer event logs through your existing logging workflow. No additional user action is required.
Fixes
Protect restricted tools when duplicating agents
When you duplicate an auto mode agent, custom tools and MCP tools that you are not allowed to add to agents are now removed from the duplicate instead of being copied. If a tool is removed, Glean shows a message asking you to reconfigure the affected steps. This helps prevent a duplicated agent from retaining tools that the new owner or editor is not permitted to use.
Analyze indexed documents when live fetch is turned off
Glean Document Reader can now use stored document bytes for indexed documents when live or federated fetching is turned off. This prevents analysis from falling back to truncated indexed snippets when the complete stored document is available, improving the reliability of document analysis for supported indexed content.
Keep uploaded files available after extraction failures
Uploaded files remain available for raw-file analysis when text extraction cannot produce usable content. This gives you another way to work with supported uploads whose extracted text is incomplete or unavailable, instead of losing access to the original file for analysis.
Make Monday MCP setup easier to complete
Monday MCP setup now explains the redirect-URL allowlisting requirement and provides guidance for contacting a Monday representative before connecting or retrying. The clearer setup path helps admins understand why a connection may fail and what must be configured before trying again.
Resume Desktop tool connections after sign-in
When connecting extension tools from the Glean Desktop app, users are now prompted to sign in if the browser is signed out. The connection continues after authentication instead of failing at the sign-in step, making it easier to authorize tools from Desktop.