# ProductAI Governance

## Run every agent safely in production.

Bring your entire agent ecosystem under one control plane. Manage access, provision credentials, enforce policies, and maintain compliance—programmatically and at scale.

[Book demo](/content/talk-to-sales/index.html) [Start free trial](https://app.guild.ai/)

**Agents**  
**47**  
**Policies**  
**12**  
**Decisions / 24h**  
**3,842**

| Agent              | Policy                     | Status       |
|-------------------|---------------------------|--------------|
| deploy-bot        | pol_deploy-gate           | Verified     |
| pr-reviewer       | pol_no-destructive        | Verified     |
| issue-triage     | pol_read-only             | Verified     |
| cost-monitor      | pol_read-only             | Verified     |
| cleanup-agent     | pol_no-destructive        | Verified     |
| docs-writer       | pol_read-only             | Verified     |
| release-notes     | pol_read-only             | Verified     |
| incident-bot      | pol_no-destructive        | Verified     |

### Policy Decisions

| Live                  | Action                    | Decision    |
|-----------------------|---------------------------|-------------|
| pr-reviewer           | createReview              | ALLOW       |
| deploy-bot            | mergePullRequest          | ALLOW       |
| cleanup-agent         | deleteRepository          | DENY        |
| issue-triage         | transitionIssue           | ALLOW       |
| docs-writer           | postMessage               | ALLOW       |
| deploy-bot            | forcePush                 | DENY        |

**Enforced pre-execution** 12 policies active

## Deliver agent control at runtime

Agents can’t be managed in silos. Guild identifies and catalogs every agent, then applies policies, permissions, and controls centrally while it runs.

### Runtime proxy

| API Call                   | Credential               |
|----------------------------|--------------------------|
| POST /repos/acme/api/merges | cred_8f2a                |
| identified                  | agt_4f19c2ba7e           |
| catalogued                  | guildai/pr-reviewer · v12 |
| governed                    | pol_31c7 · ALLOW         |
| Status                      | 200 · proxied to api.github.com |

## Know what agents are running

Shadow agents are risky. With Guild, every agent gets catalogued with its own unique ID, so you’ll always know what’s running and where.

| Agent                | Action            | Status   |
|----------------------|-------------------|----------|
| pol_31c7             | forcePush         | DENY     |
| deploy-bot → forcePushreq_9d04 | ALLOW     |
| deploy-bot → forcePushreq_9d11 | ALLOW     |
| deploy-bot → forcePushreq_9d27 | ALLOW     |

## Define per-end-point credentials

Individual agent keys are stored at the org level, granted per-endpoint, and reusable for every engineer.

### GitHub

| Action                       | Status   |
|------------------------------|----------|
| GET getPullRequest           | allowed  |
| GET listCommits              | allowed  |
| POST createReview            | allowed  |
| POST mergePullRequest        | allowed  |
| DELETE deleteRepository      | denied   |
| DELETE removeBranchProtection | denied   |
| GET listBranches             | allowed  |
| POST createIssue             | allowed  |

### Slack

| Action                    | Status   |
|--------------------------|----------|
| POST postMessage         | allowed  |
| GET listChannels         | allowed  |
| POST setTopicChannel     | allowed  |
| GET getUserProfile       | allowed  |
| DELETE deleteMessage      | denied   |
| POST kickFromChannel      | allowed  |
| GET getConversation        | allowed  |
| POST addReaction         | allowed  |

### Jira

| Action                     | Status   |
|----------------------------|----------|
| GET getIssue              | allowed  |
| POST createIssue          | allowed  |
| POST addComment           | allowed  |
| GET searchIssues          | allowed  |
| DELETE deleteIssue        | denied   |
| DELETE removeVersion      | denied   |
| GET listProjects          | allowed  |
| POST transitionIssue      | allowed  |

### Linear

| Action                      | Status   |
|-----------------------------|----------|
| GET getIssue               | allowed  |
| GET listCycles             | allowed  |
| POST createIssue           | allowed  |
| POST assignLabel           | allowed  |
| DELETE archiveProject      | denied   |
| DELETE removeTeamMember    | denied   |
| GET listWorkflows          | allowed  |
| POST updatePriority        | allowed  |

## Enforce human approval flows

Secure your agents by programmatically enforcing human approval for risky actions.

| Approval Status        | Action                 | Status     |
|------------------------|-----------------------|------------|
| 1 pending              | deploy-bot · Push to main | held by pol_31c7      |
| Resolved               | cleanup-agent · Drop staging DB | d.keane · DENIED |
| Resolved               | merge-bot · Merge release-42 | d.keane · APPROVED |
| Resolved               | security-agent · Rotate api key | a.chen · APPROVED |

## Turn audit trails into action

Guild marries governance and observability with complete runtime records to investigate issues, demonstrate compliance, and enforce agent standards at scale.

| Identity                | Agent               | Created By |
|-------------------------|---------------------|------------|
| guildai                 | deploy-bot agt_4f19c2ba7e | d.keane    |
| guildai                 | issue-triage agt_9b02e7f1c4 | d.keane    |
| guildai                 | cost-monitor agt_2d7a48910f  | a.chen     |

**3 identities** · **2 creators**  
**All verified**

## Frequently asked questions

**What is AI agent governance?**  
AI agent governance is the layer that decides what agents are allowed to do while they are running: which credentials they can use, which tools they can call, which actions require human approval, and what gets logged for audit.

**What does Guild AI Governance do?**  
Guild AI Governance catalogs every AI agent in your organization, applies policies and permissions centrally, provisions credentials without exposing raw secrets, enforces human approval for risky actions, and produces a complete audit trail for every tool call.

**How does Guild handle credentials and secrets for AI agents?**  
Agents never hold credentials directly. Secrets are stored in Guild's credential vault, scoped per endpoint, and attached only on the outbound hop, after policy is evaluated.

**Can Guild require human approval for risky agent actions?**  
Yes. You define which actions require an approval gate at the workspace or agent level (deleting data, sending customer messages, etc.).

**Which compliance frameworks does Guild AI Governance support?**  
Guild is SOC 2 Type 1 compliant, ISO 27001 certified, and GDPR compliant.

**Does Guild AI Governance work with agents built outside of Guild?**  
Yes. Guild is framework-neutral.

Transform with Guild.

See Guild in action.