Glean governance guidance

Reviewed 3 October 2026. Aicura is not affiliated with Glean.

About Glean

Glean connects to the organisation’s work apps and lets people search, ask questions and get answers across all of them, with results drawn from what each person can already open in the source systems. Its reach grows as more apps are connected and as people move from searching to chatting with the assistant, then to building agents that run on a schedule, analyse data and take actions in other systems. Good governance lets the organisation connect more of its knowledge and extend agents further while keeping control of what each one reaches and does.

Where governance makes the difference

• Information reaching wider than intended. Glean mirrors the permissions in each connected app. Where sharing in those apps is broader than anyone meant, Glean makes that information easy to find and fold into answers, including staff and employment records.
• Agents acting without clear ownership. People can build and share agents that write to other systems and run in the background. Without owners, approval and limits, an agent can keep acting long after its purpose or its builder has moved on.
• Content steering an agent. An AI agent follows instructions in what it reads, so a document or message in a connected app can influence what an agent does with the information and tools it has.

---

For the executive team

The benefit depends on getting access right first

Glean brings everything a person can open into one place. Most organisations have some sharing in their apps that is broader than it needs to be, and that matters more once a single search can reach all of it. Tidying access and checking for sensitive content before each wide rollout is what lets the organisation get the benefit without surprises.

Every shared agent needs an owner

Agents can be built by anyone given the right, shared across a department or the whole company and set to act on their own schedule. Each shared agent needs a named person accountable for it, a clear purpose and agreed limits on what it can change.

Oversight needs to keep pace with growth

New connected apps, new agents and new AI options arrive steadily. The deployment approved at launch will not be the one running a year later, so oversight works best as an ongoing practice rather than a one-off sign-off.

Questions worth putting to management

• Who is accountable for Glean, and who owns each agent shared beyond its builder?
• Which apps are connected, and what was done to check access and sensitive content before each was switched on?
• How many agents can change records in other systems, and who approved them?
• How would the organisation know if information reached the wrong people or an agent went beyond its purpose?
• If a customer, regulator or insurer asked how Glean is governed, what would be shown to them?

What good reporting looks like

Expect regular reporting on how widely Glean and its agents are used, which apps are connected, the main risks and how they are changing, any incidents or near misses and the significant decisions taken, with who made them and why.

How Aicura can help

• One place showing Glean, the apps connected to it and every shared agent, with who owns each one and what it can reach and change.
• Board reporting drafted from the governance record, covering use, risks, incidents and decisions.
• A public trust centre where customers and partners can see how the organisation governs its AI, backed by evidence a third party can verify.

---

For risk, compliance and assurance

Set where Glean sits in the risk appetite

Decide which classes of information Glean may index and which decisions its answers may inform. Answers that draw on staff records or feed decisions about people need review by someone who knows the subject before they are relied on.

Assess the platform, each connection and each agent

Assess the deployment as a whole before rollout, covering where it is hosted, which AI it uses and how permissions are kept in step with source systems. Assess each connected app before it goes live, covering what it holds and how well its sharing reflects intent. Assess each shared agent on its own terms, covering the data it reaches, the actions it can take, whether it runs unattended and who reviews its work.

Reassess when anything changes

A new connected app, a new agent that writes to other systems, a change to which AI powers the assistant or a move between hosting options all shift the deployment away from what was assessed. Each needs to trigger a fresh assessment rather than wait for the annual cycle.

Make the data and AI decisions explicitly

Decide whether to use the AI arrangement the vendor manages or to supply the organisation’s own AI accounts, and confirm the data handling and retention commitments that apply to whichever is chosen. Decide which apps and content are out of scope for indexing, how long activity records are kept, who may access them and for what purpose, and let staff know.

Confirm which protections apply

Check which hosting option, accounts and agreements carry the vendor’s enterprise commitments, and how they extend to any AI provider the organisation connects itself. Confirm the organisation’s own responsibilities under the shared responsibility arrangement, especially for identity, connector credentials and any deployment in its own cloud.

Set limits and name who acts

Agree the limits that matter, such as sensitive content found in newly connected apps, bursts of agent write activity, changes to agent schedules or a sudden shift in use, and name the person who acts when one is crossed.

Be ready to show governance to others

• Policy. Maintain an AI use policy that covers Glean specifically, including which content may be indexed, who may build, share and publish agents and which actions agents may take without a person confirming. Tie each commitment to the control that carries it out.
• Accountability. Record decisions on new connections and published agents with their reasons when they are made. Have agent owners confirm periodically that their agents still match what was approved.
• Operating evidence. Keep records that controls ran, such as sensitivity scans before each connection, access reviews, agent approvals and interventions.
• External scrutiny. Prepare for internal and external audit, certification against standards such as ISO/IEC 42001, regulator enquiries and insurer questions about AI controls.
• Customers and partners. Due diligence questionnaires increasingly ask how AI tools reach customer data. Have the Glean answer ready and supported by evidence.
• Incidents. Be able to reconstruct which agent version, which connected data, which AI and which controls were in place at the time.

How Aicura can help

• Risk and impact assessments against recognised frameworks for the platform, each connection and each agent, run again on any material change and feeding a managed set of risks and controls that can mirror to the GRC platform.
• Connections to data catalogues and privacy platforms that map which data Glean and its agents reach and raise findings where the organisation’s own data rules are not met.
• Limits on activity with responses the organisation chooses, from notifying an agent’s owner to opening an incident.
• Approvals, owner attestations and control actions kept as evidence that can be verified without relying on the organisation’s word.

---

For technology and security

Discover every form Glean takes

Inventory the web and desktop apps, every connected app, every agent shared beyond its builder and every internal application calling Glean’s APIs or its MCP server. Record which AI each agent uses and whether the deployment is vendor hosted or in the organisation’s own cloud.

Prepare access before each connection goes live

Connect apps in stages, starting with a test group. Scan new content for sensitive information before wider release, tighten sharing in the source app where it is broader than needed and keep sensitive content out of the index or hidden while it is fixed. Make sure joiners, movers and leavers flow through single sign-on and automated deprovisioning so access in Glean follows the source systems.

Put agents and connections under change control

Use moderator review so agents are approved before they are published widely, and limit which write actions may run without a person confirming, especially for agents running in the background. Where connections, data sources and agents are managed through Glean’s APIs, hold those definitions in source control and run a preflight check before each change goes live, so a new connection, agent, tool or AI is checked against agreed rules first.

Control who can build, share and publish agents

Grant agent building and publishing deliberately rather than to everyone by default, and limit how widely agents may be shared without review. Prefer pre-configured tools over custom ones for agents that write, point write actions at specific targets and keep destructive actions off until reviewed. Keep administrator roles to least privilege and review them regularly.

Govern API traffic in the request path

Calls from the organisation’s own applications and agent frameworks to Glean’s APIs pass through the organisation’s network and can be routed through an AI gateway on the way. Use it to redact personal information, screen inbound content for injected instructions, restrict which tools are called, cap spend and stop an integration when needed.

Use the network and vendor settings for what the gateway cannot reach

The assistant and agents reach the AI through Glean’s own service rather than from user devices. Traffic from managed devices to Glean can be allowed, blocked and logged at the network edge, keeping people on the approved tenant. Where the deployment runs in the organisation’s own cloud with its own AI accounts, agree with Glean whether traffic to the AI provider can pass through the organisation’s controls, and otherwise rely on the vendor’s AI settings, observation and contract.

Feed everything Glean exposes into monitoring

Bring administrator audit logs, agent run records and activity data into monitoring. Apply the limits risk has agreed, watch for subscription changes, unexpected write bursts and unusual run patterns, and set responses ranging from alerting an owner to disabling an agent.

Test agents against the live deployment

Red-team and evaluate agents that write to other systems before they are published and on a schedule afterwards, against the deployed agent with its real tools and connected data. Watch for changes in behaviour between tests, particularly after the vendor changes or retires an AI the agent relies on.

How Aicura can help

• One inventory of Glean, its connected apps, agents, AI, tools and runtime identities, with discovery of Glean use and API integrations not yet recorded.
• Preflight checks in the deployment pipeline before a connection, agent, tool or AI goes live, with every answer recorded.
• A gateway in the request path for applications calling Glean’s APIs, applying redaction, injection screening, tool limits and budgets derived from the agreed controls.
• A collector inside the network for Glean audit logs and agent run records, flagging tool calls an agent was never declared to make, plus scheduled red-teaming and evals against live agents.

---

Getting lasting value from Glean

Glean pays off when it is governed as an ongoing practice rather than approved once at launch. Three things matter whatever the role. The first is getting access right in every connected app before Glean reaches it, and keeping it right. The second is knowing every shared agent, who owns it and what it may change. The third is being able to show, at any time, how decisions were made and that controls held. Organisations that get these right can connect more of their knowledge and extend agents further with confidence.

---

Sources

• Glean, Glean Docs
• Glean, Work AI that works
• Glean, Security platform
• Glean Docs, Security principles
• Glean Docs, Secure configuration
• Glean Docs, Deployment best practices
• Glean Docs, Shared responsibility
• Glean Docs, Background agents
• Glean Docs, Agent sandbox and programmatic tool calling
• Glean Docs, LLMs
• Glean Docs, Agent access
• Glean Developers, Glean developer platform