Skip to main content

Release Notes

Welcome to the Glean Release Notes. Here, you'll find information on recently launched features and improvements.

User features

Store memories in Glean from any MCP-connected tool

Connected MCP host applications such as ChatGPT, Claude, Cursor, VS Code, and Copilot can now write memories back to Glean in addition to reading them. You can add, update, and delete memories from any MCP-connected tool, and those memories are stored in Glean with full source provenance. This means the context you build while working in your IDE or AI assistant is available everywhere Glean memory is used. Admins can enable or disable memory write operations per deployment in Admin console → Platform → Glean MCP server.

How to access: Memory write operations are enabled by default on the Glean MCP server. No additional setup is required.

Learn more

Assistant ROAD-1261


New Google and Microsoft actions

Glean Assistant can now help you complete more work across Google Workspace, Microsoft 365, Slack, and Microsoft Teams, so you can move from answers to action without switching tools. You can draft emails, create docs and spreadsheets, and send messages in the apps your team already uses, helping you finish common workflows like status updates, stakeholder communications, and tracker updates faster. Admins must first configure the relevant action packs for Google, Microsoft 365, and Slack before these options appear for users.

How to access: Administrators must enable for your organization. Admins must enable the relevant action packs in Admin console → Platform → Actions before users can access these capabilities. To support Google Docs, Google Sheets, and Gmail drafts, set up Google actions and connect the Google Drive data source. To support Outlook, Word, Excel, and Microsoft Teams actions, set up Microsoft 365 actions and link them to your Microsoft 365 data source; some Microsoft environments will require admin consent before users can connect successfully.

Learn more

Actions ROAD-1194


Update Salesforce opportunities from Glean Assistant

Sales teams can now update Salesforce opportunities directly from Glean Assistant using natural language, without switching into the Salesforce app. When a user asks Assistant to change key opportunity fields—such as amount, stage, close date, or forecast category—Assistant identifies the correct record, collects any missing details in a structured way, and writes the updates back to Salesforce while honoring the user's permissions. Assistant then confirms the change and links to the updated opportunity so sellers can review and move on to their next task faster.

How to access: Administrators must enable for your organization. To use Salesforce opportunity write actions, your organization must have the Salesforce data source connected in Glean with API access to Opportunities and related objects, using the standard Salesforce connector. In the Glean Admin console, go to Platform → Actions and ensure the Salesforce action pack and the Update Opportunity write action are enabled for your deployment. End users will need to authenticate their own Salesforce account in Glean when prompted, so Assistant can respect per-user Salesforce permissions for viewing and updating opportunities. Once configured, users can simply ask Glean Assistant to update an opportunity in plain language and confirm the proposed changes in chat.

Learn more

Actions ROAD-1156


Editing for slides and image collections

You can now edit slides and image collections by annotating exactly what you want changed and asking Glean to regenerate the result with AI. Mark up a single image or a full set, describe the change in plain language, and get updated visuals without starting from scratch.

How to access: Administrators must enable for your organization.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1402


Manually refresh interactive artifacts

You can now manually refresh interactive content in Glean to update the underlying data while keeping the existing layout and design intact. This is especially useful for dashboards and other data-driven pages that need fresh information without rebuilding the page from scratch. Each refresh creates a new version and shows when the content was last refreshed.

How to access: Administrators must enable for your organization.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1343


Glean Assistant Meeting Notes

Glean Assistant Meeting Notes captures and transcribes your meetings, then generates an AI summary with key decisions and action items after the meeting. Transcripts and summaries are indexed in Glean and surfaceable through search and chat. It's designed to reduce the manual work of note-taking, speed up follow-ups, and help teams get answers to difficult questions without losing momentum in the conversation.

This is a desktop Mac-only feature available for Glean Key customers. Admins can control the rollout of this feature to their end users from Admin console → Assistant → Meeting Notes. The default setting lets Glean manage the rollout: "Meeting Notes" is off by default and will be turned on for users no earlier than June 8, 2026. We recommend starting with "On only for admins"; if you like it, you can turn it on for everyone in your organization at any time, or select "Off" to opt out. Your choice can be changed at any time and will not be overwritten. This feature may use credits. Learn more in Glean's Help Center.

How to access: Administrators must enable for your organization. We enabled admin configuration on May 7, 2026. Admins will be able to opt in.

We'll notify admins on May 20, 2026 about the feature and that they can opt in. GA and billing start for customers who opt in starting May 21, 2026.

We will default enable anyone who did not disable during the notice period on June 8, 2026.

This applies only to desktop users on Mac.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-769


Azure DevOps repository, pull requests, and code indexing

Glean now indexes Azure DevOps repositories, code files, commits, and pull requests, so engineering teams can search code and use Assistant to find relevant files, recent changes, and PR context in one place. This also adds permission-aware crawling for Azure DevOps code content using native project and repository ACLs.

How to access: Administrators must enable for your organization. In the Glean Admin console, open your Azure DevOps data source and review any setup warnings or validation alerts to make sure the service principal has been granted access to repositories. If you use Microsoft 365 identity mapping for Azure DevOps permissions, make sure your O365 connector is configured and healthy so indexed content appears correctly in search.

Learn more

Connectors ROAD-1246


Glean Assistant can now query Snowflake with natural language

Glean Assistant can now query Snowflake directly so users can ask questions in natural language or run SQL against approved projects and datasets, without leaving Assistant. This makes Glean a front door to your Snowflake data warehouse so business users get governed, self-serve access to live data while data teams stay in control of what is exposed and how it is queried.

How to access: Administrators must enable for your organization.

Learn more

Connectors ROAD-1166


Claude Opus 4.8 in Assistant and Agents

Claude Opus 4.8 is now available in Glean Assistant and Agents. Opus 4.8 is Anthropic's latest premium model for tasks involving code generation and agentic workflows.

How to access: Glean-hosted customers can select Claude Opus 4.8 directly in Assistant and Agents where model choice is enabled. If your organization manages model availability centrally, admins can make Claude Opus 4.8 available to end users and agent builders through model configuration in Glean.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Agents ROAD-1465


Gemini 3.5 Flash in Assistant and Agents

Gemini 3.5 Flash is now available in Glean Assistant model choice and in the Agents model hub. Gemini 3.5 Flash offers significantly better cost efficiency than larger frontier models while maintaining strong reasoning for well-scoped agent tasks.

How to access: Glean Universal Model Key: Gemini 3.5 Flash appears automatically in the Assistant model choice dropdown and Agents model hub. Admins can exclude it via Admin console → Platform → Models.

Customer Key: Gemini 3.5 Flash is available via Admin console → Platform → Models.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Agents ROAD-1445


Persistent permissions for tool use

You can now control exactly how Glean uses your connected tools with persistent permissions. Set any action to "Always Allow," "Ask," or "Block" — and your preference sticks across every session, whether you are in Assistant, Agents, or MCP. A new Connectors pane in your settings gives you a single place to see which apps you have connected, manage authentication, and fine-tune permissions for each tool individually.

Learn more

Agents ROAD-1368


Dictation in Assistant composer

You can now use your voice to compose messages in Glean Assistant. Dictation is back with a redesigned experience and a significantly upgraded transcription model supporting 98 languages — with high-accuracy results for dozens including English, Spanish, Chinese, and many more. Tap the microphone icon in the composer on web, desktop, or mobile, speak naturally, and your words appear as editable text. Perfect for capturing long thoughts, drafting on the go, or working in your preferred language.

Learn more

Assistant ROAD-1456


Greenlist application URLs for functional buttons in HTML artifacts

Interactive artifacts can now include functional buttons that link directly to your work applications — like Google Docs, Salesforce, Jira, or any other approved application. When you generate an interactive page with action buttons, those buttons actually take you to the tool where you need to act. Non-approved applications will still be accessible after a user clicks through a confirmation modal.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1437


Save chats started from the Glean browser extension sidebar in chat history

Chats started from the Glean browser extension sidebar can now be saved in chat history, making it easier to revisit in-context conversations across sessions. This reduces repeated questions and gives users a more consistent experience between the extension and the main Glean chat experience. It is intended for end users of the browser extension.

Learn more

Assistant ROAD-1317


Build, edit, and share spreadsheets in Glean Assistant

Create and work with spreadsheets directly in Glean canvas. Upload a spreadsheet or ask Glean to generate one from scratch, then sort, filter, edit cells, and refine it through follow-up prompts without leaving the conversation. You can also export your sheet to Excel or copy it into other tools.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1300


Support zip file analysis in Glean Assistant

You can now upload and analyze .zip files directly in Glean Assistant in thinking mode. Whether you're working with bundled documents, code packages, or compressed datasets, just drag and drop your zip file into the composer and start asking questions about its contents.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1268


See MCP usage and adoption in Insights

Admins can now see MCP usage insights in Glean, making it easier to understand adoption and activity across MCP-connected experiences. This gives admins clearer visibility into how people are using MCP, including usage metrics that teams had previously been tracking outside the product.

Learn more

Embedded Integrations ROAD-1126

Admin features

LLM admin refresh for Glean Key customers

Admins get a clearer, more precise way to manage model availability for Glean Key deployments, including per-model toggles and surface-specific restrictions for Assistant and Agents. This improves governance for model access and makes the admin experience easier to understand and operate.

Learn more

Admin Capabilities ROAD-1404


MCP insights

Admins can now see how their organization is using Glean MCP directly in Insights, with visibility into adoption and activity across connected MCP hosts and tools. This helps teams understand which apps are driving value, identify active users, and track MCP usage without relying on manual reporting or backend logs. This feature is for admins only.

How to access: Open the Insights page in the Glean Admin experience to view MCP usage and adoption data. No additional setup is required beyond having Glean MCP enabled for your deployment. If MCP has not been enabled yet, first turn on the Glean OAuth Authorization Server under Users and permissions → Third-party access (OAuth), then enable Glean MCP server under Platform → Glean MCP server.

Learn more

Admin Capabilities ROAD-1329


Turn slide generation default on for all users

Slide generation in Assistant will be enabled by default for all customers no earlier than June 8, 2026. To keep it off or limit access, select "Off" or "On only for admins". Your choice will be respected, and you can change it anytime. If you have already enabled slide generation "On", no action is needed.

If your company has a slide template configured, Glean will use it automatically. If not, Glean will take styling inspiration from your company's existing visual assets so decks still feel on-brand.

How to access: Admins can access this toggle by navigating to Admin console → Assistant → Settings → Slide deck generation.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1377

Admin Features

Merge Microsoft 365 connector set up

Admins can now set up Microsoft 365 connectors for SharePoint and OneDrive through a shared Microsoft 365 setup flow, which reduces duplicate configuration and makes setup easier to manage.

This change is for admins only and currently applies to SharePoint and OneDrive only — Outlook and Teams are not included in this launch.

The new flow introduces shared/common settings at the Microsoft 365 level, while SharePoint- and OneDrive-specific configuration remains in their respective child setup flows.

Admins should be prepared to complete the Microsoft 365 parent setup first, then finish any SharePoint- or OneDrive-specific permissions and settings needed for each connector.

How to access: In the Admin Console, start with the new Microsoft 365 connector setup, which now acts as the shared parent setup for SharePoint and OneDrive.

After completing the Microsoft 365 setup, finish configuration for SharePoint and/or OneDrive in their child setup flows, including any connector-specific permissions, domains, certificates, or REST API setup required for your environment.

If you already use SharePoint or OneDrive, review the updated setup documentation because credentials and shared authentication are now handled through the Microsoft 365 parent flow rather than as fully separate standalone setups.

Learn more

Admin Capabilities [ROAD-997]


Sensitive content models: Find the signal in the noise

Glean’s sensitive content models separate the signal from the noise by pairing traditional infotype detection with AI models trained on your full enterprise context, such as document content and permissions, to surface true exposure. Each finding receives a low/medium/high severity label with plain-language explanations. Our sensitive content models reduce false positives and achieve about 80% accuracy rate on unstructured data. Both GCP and AWS are supported.

How to access: Available in Glean Protect+. Super admins or Sensitive Content Moderators can go to Admin Console → Glean Protect → Sensitive findings → Policies, then create or edit a policy and select Enable document analysis.

Learn more

Protect [ROAD-753]

User Features

Auto mode agents powered by natural language

Auto Mode agents are now generally available, featuring a redesigned natural language agent builder where you simply describe what you want, answer a few guided questions, and reach a testable first draft faster than ever — with full control to refine as you iterate. This release also introduces debug and trace views for step-by-step visibility into agent behavior, and an agent sandbox for dynamic context management and code execution on complex tasks. From idea to production-ready agent, faster — with more power, transparency, and flexibility built in from the start.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

How to access: Administrators must enable for your organization.

Learn more

Agents [ROAD-730]


Nested Container Search in Chat/Agents

Chat and Agents can now search within tagged folders, spaces, sites, and other containers more effectively, including content nested inside subfolders. This helps users get more complete, relevant answers from large containers without having to sift through top-level contents first, and improves performance by focusing only on matching documents. This applies to Assistant (Chat with AL), Agents Plan and Execute, Agents Autonomous mode, and the Agents Company Search step. It does not change Chat V2 or workflow agents.

How to access: Once this feature is available in your environment, users can start tagging supported container URLs in Chat or using supported container sources in eligible Agent steps. If your team manages custom search configuration, make sure co.gsto.enable_container_filter is enabled. No other admin setup is required.

Learn more

Agents [ROAD-1362]


View images in Assistant and Agents for custom datasources

Assistant and Agents can now caption, index, and render images from content ingested through custom data sources, not just native connectors. This helps people get more complete answers from proprietary document repositories when diagrams, charts, screenshots, or other visuals are important to understanding the source content.

How to access: Administrators must enable for your organization.

Learn more

Agents [ROAD-1314]


Model Choice in Assistant for Customer Key

Customers on their own LLM key now have access to model choice drop down in Glean Assistant where they can select from a variety of models their organization has enabled.

How to access: Administrators must enable for your organization.

Learn more

Assistant [ROAD-1288]


Blocked Chat Status Indicator

Chat status indicators make it easier to keep work moving in Glean by showing which conversations need your attention, including chats with new updates, chats that are ready for review, and chats where an agent is waiting on your confirmation or additional input. This helps users quickly return to long-running or asynchronous work when results are ready.

Learn more

Assistant [ROAD-1367]


See step-by-step elapsed time in Assistant plans

Glean Assistant now shows the elapsed time for each step while it works through a multi-step plan, so you can better understand progress and how long each part of a task takes. This makes longer-running work more transparent and helps you decide whether to wait, redirect, or refine the request. This is available to all end users in Assistant when a plan is shown for complex, multi-step queries in Thinking mode.

How to access: No setup is required. The elapsed time appears automatically in Glean Assistant when a complex query triggers a plan in Thinking mode.

Learn more

Assistant [ROAD-1327]


Sigma connector

Connect Sigma to Glean so users can search and discover Sigma dashboards, workbook pages, and chart elements directly in Glean. This helps teams get to the right BI content faster, brings more analytics context into Glean, and respects Sigma permissions.

How to access: Administrators must enable for your organization.

Learn more

Connectors [ROAD-1192]


Addition of boards and work items to Azure Devops connector

Glean now supports indexing for Azure DevOps boards and work items, including Epics, Features, Stories, Tasks, and Bugs. This enhancement enables users to search for project status and work items from Azure DevOps directly within Glean, streamlining project tracking.

How to access: Administrators must enable for your organization.

Learn more

Connectors [ROAD-914]


Glean app in ChatGPT

Automatically connect to Glean within ChatGPT apps with a single click of a button. You can get full enterprise context in OpenAI ChatGPT.

How to access: Admins must first configure the Glean MCP connector in ChatGPT and enable both Glean OAuth and Glean MCP servers in the Glean Admin Console. After setup, end users can open a ChatGPT conversation, click the plus icon, open More, and select Glean to activate it for that conversation.

Learn more

Embedded Integrations [ROAD-1111]

Admin Features

Add Glean usage logging across GCE and Insights

Admins can now track Glean usage more consistently across Glean Customer Event Logs and Insights, making it easier to audit activity, analyze adoption, and answer questions about who used Glean, when, and where. Glean Customer Event Logs are designed to capture key user actions for analytics and reporting, including searches, chats, feedback, and workflow activity. This improves visibility for reporting, debugging, and downstream analysis across Glean surfaces.

How to access: Admins can use Glean Customer Event Logs to analyze usage activity and build custom reporting workflows. By default, these logs are written as a unified glean-customer-event stream to storage in your Glean project, and are typically loaded into a data warehouse such as BigQuery or Athena for analysis. Admins can also use Insights in the Admin console to review adoption and usage trends, including the Overview tab and Insights chat for natural-language analysis of platform usage.

Learn more

Admin Capabilities [ROAD-1281]


Add IdP-level permissions for Agent Scheduling

Admins can now control access to Agent Scheduling with identity provider groups, not just individual users. In Admin console → Platform → Agents → Scheduled triggers, you can add Google Groups or Azure AD / Entra ID groups alongside individual users, so the right teams can schedule agents without manual user-by-user setup.

How to access: To use this feature, go to Admin console → Platform → Agents → Scheduled triggers and turn scheduled triggers on for everyone or some users. If you choose a limited rollout, add individual users or supported IdP groups from Google Groups or Azure AD / Entra ID to the access list; members of those groups will inherit scheduling access automatically.

If your scheduled agents need to send outputs automatically, also configure the relevant action packs to allow Run without user confirmation for agents; otherwise, background runs can't use those actions.

Learn more

Admin Capabilities [ROAD-1219]


Model Insights for Customer Key Admins (BYOK)

Customer Key admins can now use LLM Insights in Glean to monitor Glean-attributed LLM usage and reliability, helping them provision capacity and troubleshoot issues with more confidence. The dashboard shows key metrics like peak and average TPM, peak RPM, 429 rate, and 5xx rate, with model-level breakdowns so admins can distinguish quota pressure from provider instability faster.

This release is available only for Customer Key (BYOK) deployments in v1, and it is not available for Glean Key deployments. Data updates once daily, may be delayed by up to 24 hours, and is intended for trend analysis and capacity planning rather than real-time monitoring or billing reconciliation.

Learn more

Assistant [ROAD-1280]


Glean Agentic Search Model [Glean Key, Glean Hosted Only]

Glean's Agentic Search Model is a retrieval-optimized model that runs before the frontier model on eligible Assistant queries. It first gathers the most relevant enterprise context, then passes that evidence to the frontier model so responses are faster, better grounded, and more efficient.

In testing, this architecture delivered roughly 50% lower latency with no regression in answer quality or satisfaction.

Available for customers hosted by Glean on GCP using the Glean Universal Model Key. Not applicable to AWS, Azure, customer-managed key deployments, or EU customers.

Learn more

Assistant [ROAD-1250]


Agent alignment check for write actions

Glean Protect now has an agent alignment check before agent write actions run, helping detect unsafe or misaligned actions before they can make changes in connected systems. This strengthens protection for higher-risk agent workflows, especially background and automatically triggered agents, where write actions can run without per-run approval. When configured in flag-for-review mode, potential violations are logged in the Findings dashboard so admins can investigate without interrupting the run.

How to access: Admins can manage AI security policies in the Admin console under Glean Protect → AI security → Policies. From there, they can choose which agent types a policy applies to, including Automatically triggered agents, and set the enforcement action to Flag for review so issues appear in the Findings tab for triage.

Learn more

Protect [ROAD-811]

User Features

In-line previews for write actions in Assistant

Assistant now shows a richer in-line preview before a write action runs, so users can quickly understand what will happen before they create, update, or post anything. The preview surfaces key parameters in a compact, human-readable card and uses a clearer action button such as Create, Update, or Post, which makes actions feel more transparent, predictable, and safer to use without breaking conversation flow.

How to access: Write actions in Glean let users complete tasks like creating tickets, updating records, sending messages, or exporting content directly from a conversation. In the Glean web app, write actions can pause for review before they run, and for supported actions users can review and edit fields or content inline before saving the change. Access depends on your organization having the relevant actions enabled and authenticated in the Glean Admin console, and actions always respect the user's existing permissions in the connected application.

Learn more

Actions [ROAD-1193]


Glean agents support Snowflake Cortex Agents API

Glean is introducing new functionality enabling Glean agents to call Snowflake Cortex agents to answer natural language questions over Snowflake‑governed data. The integration is powered by Snowflake's Cortex Agents REST API, so customers can reuse their existing agent orchestration, tools, and semantic models directly inside Glean.

How to access: Administrators must enable for your organization.

Learn more

Actions [ROAD-1056]


See contextual images in Glean Assistant responses (AWS customers)

Glean Assistant now displays relevant images from Google Drive and SharePoint documents alongside text responses when appropriate. This enhancement helps users better understand complex information by providing visual context directly in chat answers—no extra steps required. Images are shown only when relevant to the user's query, with full permissions enforced for every user.

This was previously released for GCP customers.

How to access: Administrators must enable for your organization.

Learn more

Agents [ROAD-916]


GPT 5.5 in Assistant and Agents

Glean now supports OpenAI GPT-5.5 in Assistant and Agents as a premium model. In Assistant, users can select GPT-5.5 from model choice. In Agents, builders can use GPT-5.5 in the Model Hub for supported agents and steps. GPT-5.5 is available for Glean Universal Model Key deployments and for Customer Key deployments through OpenAI, and premium model usage may consume FlexCredits or other usage-based charges based on your plan.

How to access: Administrators must enable for your organization. In the Admin console, go to Platform → LLMs and make sure GPT-5.5 is enabled in the Model Hub. For Assistant, users can then select GPT-5.5 from the model picker when model choice is available. For Agents, builders can choose GPT-5.5 as the default model in agent settings or for individual eligible steps in the canvas.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Agents [ROAD-1316]


Write tool previews, batch writes, and simplified auth in Assistant

Write tools in Glean now show a clear preview before they run, so users can review and approve changes before execution. This applies to both single-item and batch write flows. We also simplified tool auth to reduce setup friction for admins and end users.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Agents [ROAD-1299]


Additional SharePoint List support

Glean now supports indexing four additional types of SharePoint Lists: Discussion Board, Custom List in Datasheet View, Tasks, and Issue Tracking. This expands search coverage for content teams commonly use to manage discussions, track work, and organize operational data, making it easier for users to find more of their SharePoint knowledge in Glean.

How to access: Administrators must enable for your organization.

Learn more

Connectors [ROAD-1211]


Switch models within an existing Assistant chat session

You can now switch models mid-conversation in Glean Assistant without losing your thread. Previously, changing models would drop you into a new chat — losing your context, canvas, and conversation history. Now your session stays intact, so you can start with a faster model and upgrade to a stronger one when you need it, or experiment across models without starting over.

Learn more

Assistant [ROAD-1315]


Updated navigation, chat bar, and personalized activity cards in Assistant

Glean has been updated with a more streamlined layout, including a refreshed chat bar and new proactive cards that surface relevant next steps based on the context of your work. The updated navigation introduces a dedicated Search entry point along with a new content library. Together, these updates create a more unified way to search, chat, and work within Glean. Most users will see the change on April 29, 2026.

Customers who were part of the earlier gradual rollout of the new layout (ROAD-1040) will now be moved to 100% rollout on April 29, 2026.

Learn more

Assistant [ROAD-1318]-MR


Experience real-time, hands-free voice conversation support with Glean on web, desktop, and mobile

Experience natural, hands-free real-time voice conversations with Glean Assistant on web, desktop, and mobile. Talk to Assistant and hear it respond immediately and naturally, with low-latency, back-and-forth dialogue—just speak, and Glean responds out loud, allowing you to interrupt or steer the conversation at any time. Real-time voice unlocks new ways to brainstorm, prepare for your day, and get work done wherever you are, making Glean more accessible and productive for users on the go or at their desk.

This feature is available for Glean Key customers. Admins can control the rollout of this feature to their end users from Admin console → Assistant → Realtime Voice. The default setting lets Glean manage the rollout: voice is off by default and will be turned on for users no earlier than mid-April 2026. We recommend starting with "On only for admins"; if you like it, you can turn it on for everyone in your organization at any time, or select "Off" to opt out. Your choice can be changed at any time and will not be overwritten. This feature may use credits. Learn more in Glean's Help Center.

How to access: Administrators must enable for your organization.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant [ROAD-911]


Deprecation of Claude Sonnet 4, Sonnet 4.5 and Opus 4.5

Anthropic is retiring the 1M token context window beta for Claude Sonnet 4.5 and Claude Sonnet 4 on April 30, 2026. After this date, the context-1m-2025-08-07 beta header will have no effect and requests exceeding the standard 200k-token context window will return an error.

Users should migrate to Claude Sonnet 4.6 or Claude Opus 4.6, which support 1M context at standard pricing with no beta header required.

Ref: Anthropic release notes

Learn more

Agents [ROAD-1271]


See live task progress in Assistant plans

Glean Assistant now shows a live task list for complex, multi-step requests, so you can see the plan it creates and track progress as work happens. Each step updates in real time as Pending, In Progress, or Done, which makes longer-running research, analysis, and workflow tasks easier to follow and less opaque. You can also stop Assistant and redirect it if you want to change course mid-task. This is available in Thinking mode for all users on Glean Universal Key and Glean Hosted.

How to access: Open Glean Assistant and submit a complex request in Thinking mode, such as a multi-step research or analysis task. If Assistant decides the request needs planning, it will show a task list card in chat with live step-by-step progress. No admin setup is required.

Learn more

Assistant [ROAD-1332]


Create images with Open AI's Image 2 model in Assistant and Agents

Glean now supports OpenAI Image 2, giving users access to a new, high-quality image generation capability from OpenAI that was not previously available. Users can generate richer visuals directly from Glean Assistant and Agents, with images embedded into artifacts like slides and docs where supported. Agent builders also now have a dedicated Image Creation node, and Plan & Execute plus autonomous agents can explicitly incorporate image generation into multi-step workflows. Usage continues to follow your organization's existing safety, quota, and eligibility settings, with graceful fallbacks to text-only responses when requests are blocked or exceed limits.

This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant [ROAD-1324]


See file pills for Memory and Skills in intermediate steps

Glean Assistant now shows file pills for triggered skills in intermediate steps, so users can more easily see which skill influenced a response and open it in the side panel for more context. This makes Assistant less opaque during longer-running tasks and gives users a clearer view into how responses are being shaped. No admin setup is required.

How to access: When Assistant shows intermediate steps during a query, look for the skill pill in the progress area and click it to open the skill in the side panel.

Learn more

Assistant [ROAD-1323]


Import Memory to Glean from other AI Providers

You can now bring work-related preferences, role context, active projects, and communication habits from other AI assistants into Glean, so Assistant can personalize responses faster without making you start from scratch. Users can open Settings, go to Personalization, copy Glean's import prompt, paste it into another AI assistant, then paste the response back into Glean to save imported memories. Imported memories are reviewed and written into Glean's memory system, where users can later view or delete them from Personalization. Fast mode is not supported for this flow.

How to access: Open Glean, go to Settings, and select Personalization. Use the import option to copy Glean's prompt into another AI assistant, then paste that assistant's response back into Glean to add work-related memories. No admin setup is required beyond having memory enabled for your organization.

Learn more

Assistant [ROAD-1304]


Agent Sandbox & Programmatic Tool Calling [Customer-Hosted on GCP, AWS]

Agent sandbox and programmatic tool calling are now available for customer-hosted deployments in AWS and GCP. In Thinking mode, Glean Assistant can perform secure code-backed analysis, large result retrieval, and multi-step reasoning while keeping execution and data inside the customer's cloud environment. Programmatic tool calling lets Glean orchestrate large numbers of tools for parallel research tasks by writing and executing code in a sandbox.

How to access: After setup, Glean Assistant Thinking mode automatically activates sandbox when a query benefits from code execution, exhaustive retrieval, or structured analysis. For best results, use a supported frontier model such as GPT 5.4 or later, Claude Sonnet 4.6, Claude Opus 4.6 or later, or Gemini 3.1 Pro.

Learn more

Assistant [ROAD-1285]


Add, find, and revisit content in the library

The library is a new dedicated tab for everything you've created and saved in Glean. It replaces the Content section in the left navigation with a single, searchable surface for all your documents, slides, images, emails, interactive pages, Collections, Pins, and Go Links. You can also discover content that teammates have shared with you or your organization.

How it works: Open the library from the left navigation to browse everything in one place. Filter by type to narrow results, or search to find a specific item. Clicking any item opens it in context with its original conversation.

Content created with Assistant is retained indefinitely, until a user deletes it from the library. This content includes artifacts such as documents, slides, emails, and images.

This does not apply to standard chat responses, where your organization's chat retention period still applies. The library help page on docs.glean.com will include a full description of content types when the library reaches GA.

Learn more

Assistant [ROAD-1278]


Allow users to manually rename chat conversations

Users can now manually rename their chat conversations in Glean Assistant by clicking on any chat title and editing it inline. Custom titles persist across sessions, making it easy to organize and revisit past conversations. Chats you don't rename will continue to use automatically generated titles, so no action is required.

Learn more

Assistant [ROAD-1210]

User Features

One click install of Glean's remote MCP server to claude code, cursor, desktop apps

Use your MDM provider (Jamf, Kandji, Intune) to configure Glean's remote MCP server on your desktop apps (Claude Code, Codex, Cursor, Gemini CLI, Goose, JetBrains AI assistant, Visual Studio Code, Windsurf).

How to access: Admins can now deploy a Glean MCP remote server to user devices through their organization's MDM server, so users do not need to manually set up Glean MCP themselves.

Learn more

Admin Capabilities ROAD-1123


Claude Opus 4.7 in Assistant and Agents

Opus 4.7 is now available in Glean Assistant and Agents! This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Agents ROAD-1296


Glean Assistant can create Google Slides presentations

Glean Assistant can now generate full slide decks with day zero support for Google Slides presentations directly from prompts. Ask Assistant to create a slide deck on any topic, and it will automatically use your company's approved template to maintain consistent branding and formatting across all presentations. Available for Glean Key and Customer Key. Admins must add their company slide template to the admin console via Console > Assistant > Settings > Slide deck generation > upload template. This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

How to access: Admins must enable Google slide generation via the Admin console for admins or all users, and upload a company template.

Learn more

Assistant ROAD-1272


View changes between versions of your documents, emails, and messages

You can now see exactly what changed between versions of a document, email, or message generated in Glean Assistant. When you ask the assistant to revise an artifact, the diff view highlights additions, deletions, and edits inline — so you can review changes at a glance instead of re-reading the whole thing. Whether you are iterating on a customer email, refining a proposal, or editing a team update, diffs make every revision transparent and easy to approve.

Learn more

Assistant ROAD-1305


Experience refreshed progress updates

We've improved the intermediate steps experience in Assistant so it's easier to scan tool execution and reasoning while a task is running. This makes progress clearer, helps users better understand what Assistant is doing, and creates a more polished overall experience.

How to access: The intermediate steps automatically invoke for queries in assistant.

Learn more

Assistant ROAD-1294


Ask clarifying questions in Glean Assistant

Glean Assistant can now ask concise, contextual clarifying questions when a request is ambiguous or under-specified, helping it choose the right path before taking action. This improves first-pass quality, reduces wasted work caused by incorrect assumptions, and makes Assistant feel more collaborative on multi-step tasks.

How to access: Clarifying Questions activate automatically for Glean Assistant.

Learn more

Assistant ROAD-1269


MCP Apps Extension support in Glean Assistant

You can now bring your favorite applications right into Glean with MCP App Extensions. Glean acts as an MCP app host, so when an MCP server returns a UI resource, the app can render directly inside the chat experience as an interactive widget.

That means users can interact with dashboards, forms, approval flows, and data visualizations natively in Glean—without switching tools or opening a separate window. All third-party UI runs in a secure, sandboxed environment with enforced content security policies, so you get a seamless experience without compromising safety. This feature may be subject to usage-based pricing. Learn more in Glean's Help Center.

Learn more

Assistant ROAD-1234


Salesforce object query filters in setup page

Admins can now add query filters to Salesforce objects directly in the Glean Admin Console, so you can control which Salesforce records are indexed without backend configuration changes. Filters support SOQL-style conditions and combine multiple rules with AND logic, so only records that match all configured conditions are indexed. This makes it easier to limit crawl scope, support compliance and data minimization needs, and manage Salesforce indexing in a more self-serve way.

How to access: In the Admin Console, go to Data sources → Salesforce → Objects, then add a new object or edit an existing one. Open Advanced options, select Add filter, choose the Salesforce field, operator, and value for each condition, then save your changes and run a crawl for the filters to take effect. Filters apply only to the object you configure, so if you need similar scoping on related objects, configure filters separately for each one.

Learn more

Connectors ROAD-1265


Native search replacement in Atlassian's Confluence and Jira is moving to the browser extension

Glean's Atlassian Native Search replacement previously available within Jira and Confluence will revert to Atlassian's native search capabilities over the coming weeks. However, users can still search from Atlassian pages with the Glean browser extension sidebar, which lets them leverage Glean search without leaving the browser or the Atlassian page. This change is being made due to low usage and the broader benefits and extensibility of the Glean extension.

Learn more

Search ROAD-1276

Admin Features

Glean now supports custom static & dynamic headers for MCP

Glean now supports customer-specific custom HTTP headers for outbound requests to external MCP servers, making it easier to connect servers that require tenant-specific metadata, vendor-specific headers, or request-time identity context. Customers can configure any fixed static headers and dynamic headers for user email, user ID, and API key.

How to access: Reach out to your support representative with the list of headers you want to enable for MCPs and they can update your config.

Learn more

Actions ROAD-1306


Simplified OAuth management for admins

Admins can now manage OAuth settings in a clearer, more organized Third-party access page. The updated experience separates IDP-configured OAuth and the Glean OAuth Authorization Server into distinct views, making it easier to find the right settings and manage third-party access with more confidence.

How to access: Go to the Admin Console and open Settings → Third-party access (OAuth) to manage OAuth configuration. From there, admins can enable the Glean OAuth Authorization Server and create static OAuth clients with the redirect URIs and scopes needed for their internal or third-party applications. If you're using MCP, enable the OAuth authorization server first, then enable Glean MCP Servers under Platform → Glean MCP servers.

Learn more

Admin Capabilities ROAD-1216


Latency enhancements for Egnyte customers with multiple admin accounts

Glean's Egnyte connector now supports multiple administrator authentications for a single Egnyte data source, so you can authorize more than one Egnyte admin account. Glean uses every active admin token during Egnyte crawls, which improves crawl throughput as you add admins while continuing to enforce Egnyte's native permissions. This helps large Egnyte deployments index content faster and keep more content fresh in Glean.

Learn more

Connectors ROAD-1238


OneDrive and SharePoint connector enhancements

Glean supports real-time search across Microsoft OneDrive and SharePoint, including Document Libraries, Sites, and Lists. The feature streamlines collaboration and ensures users can quickly find and use the most up-to-date documents across their Microsoft 365 environment.

How to access: Admins can turn on the feature in the Glean Admin Console. For OneDrive and SharePoint, setup requires registering an app in Azure, granting the necessary application permissions (including Sites.FullControl.All and Files.ReadWrite.All), and completing the configuration in Glean. Once enabled, users can paste OneDrive or SharePoint links into Assistant to fetch and interact with documents, provided they have access rights. For detailed setup steps, refer to the Glean Help Center documentation for OneDrive and SharePoint connectors.

Learn more — OneDrive | Learn more — SharePoint

Connectors ROAD-1141


NetSuite connector

Glean now offers a NetSuite connector, enabling organizations to securely search key financial and operational data from their NetSuite ERP. The connector supports high-value workflows such as viewing invoices, payment histories, unpaid invoices, and customer records directly within Glean. By making critical NetSuite data discoverable, teams can streamline finance, sales operations, and back-office processes, improving visibility and productivity across the business.

Learn more

Connectors ROAD-842


Glean OAuth - Dynamic client registration

Glean now supports a generally available OAuth authorization server for secure third-party access and MCP host connections. Admins can enable Dynamic Client Registration so supported MCP hosts can register at runtime, apply redirect URI restrictions for tighter control, and manage connected OAuth applications, including revocation, from the Admin Console. This makes it faster to connect approved tools to Glean while giving admins more visibility and control over access.

How to access: In the Admin Console, go to Settings > Third-party access (OAuth) and enable the Glean OAuth Authorization Server. If you want to control which dynamically registered apps can connect, open Manage settings, turn on Restrict Dynamic Clients by Redirect URI, and use either the Glean-managed list or your own static list of allowed redirect URI patterns. If you're using Glean MCP Servers, also go to Platform > Glean MCP servers and enable them there.

Learn more

Developer Experience ROAD-1205


Restricted topics policies (acceptable use policies / topic guardrails) on user prompt for Assistant compliance

Glean now enables organizations to enforce Acceptable Use Policies (AUP) at the user query level in Glean Assistant and Agents. Admins can define and centrally manage topic-based guardrails to block or flag user prompts on sensitive topics—such as performance reviews, disciplinary actions, compensation, and other HR-sensitive areas—ensuring compliance with internal policies and regulatory requirements. This feature helps reduce the risk of information leakage, supports adoption in regulated industries, and builds trust by preventing prohibited queries from being processed or answered, regardless of underlying data source permissions.

Learn more

Protect ROAD-872