Guided agent authoring
Configure instructions, knowledge files, model aliases, tools and example prompts through a guided builder.
Agent Fleet
Your agents.
One governance model.
Build and host agents with approved models and tools. Discover and review Azure AI Foundry agents alongside them. Keep ownership, approvals and policy evidence in one fleet.
Governed agent publishingGoverned
-
01
Define the agentInstructions, context and examples
-
02
Choose approved resourcesModel alias, tools and toolsets
-
03
Test and submitDraft, review and publishing workflow
-
04
Operate with ownershipAccess, sessions and usage visibility
Identity · Policy · EvidenceWorkflow illustration
From useful idea to approved agent
Let business teams build.
Keep governance connected.
Approved tools and models
Use LLM aliases and MCP servers from the governed registries. Select curated toolsets and configure who can access the agent.
Review before publishing
Save drafts and submit agents for approval. Track the publishing workflow and business ownership in the control plane.
Available today
Build here.
Govern beyond.
Bring hosted agents and discovered external agents into a shared review process. Each integration has a clear operating boundary.
Hosted & governed
AI Warden agents
Build agents with the integrated Agent Builder. AI Warden hosts their runtime, manages access and publishing approvals, and applies configured gateway controls to their model and MCP traffic.
Discovery & governance
Azure AI Foundry
Discover agents through the Azure connector. Review ownership, approvals, configured tools, guardrail evidence and policy gaps, with token telemetry where Azure exposes it.
Azure remains the runtime owner. AI Warden records governance decisions locally; the discovery connector is read-only. Inline enforcement applies to traffic routed through the gateway.
A business use case
Turn repeatable work
into a governed service.
Define a narrow job
A travel and expense assistant can help staff interpret approved policy documents. Clear instructions and knowledge define its intended scope.
Choose the right boundaries
Select an approved model alias, attach approved context and grant only the tools and access groups the workflow needs.
Review before release
Test representative questions, keep the agent as a draft while iterating and submit it for publishing approval when ready.
Keep operations visible
Review supported session and usage data alongside the gateway decisions produced by model and MCP traffic routed through AI Warden.
Planned for future releases
Your agent ecosystem
will keep growing. So will ours.
We plan to extend agent discovery and governance to additional enterprise platforms, helping teams review their expanding agent estate through AI Warden.
Planned integration
Snowflake
Extend agent discovery and governance into your data-platform ecosystem.
Planned integration
Atlassian
Bring agents in collaboration and work-management platforms into governance review.
Planned integration
Salesforce
Expand agent discovery and governance across customer and business workflows.
These integrations are planned for future releases. Availability and scope may change, with more platforms to follow. Discuss your integration priorities.
Questions, answered
A clearer path to adoption.
Does every agent have to be built in AI Warden?
No. Agent Fleet includes AI Warden-hosted agents and Azure AI Foundry agents discovered through the Azure connector. The integrated Agent Builder is available for authoring your own agents. External discovery depends on connector permissions; inline controls depend on traffic using the gateway.
Can teams bring their own knowledge and tools?
Agents can use approved context files, model aliases, toolsets and registered MCP servers. Access and publishing decisions stay within the configured governance workflow.
See AI Warden in action
Know your agents.
Make their boundaries explicit.
See hosted-agent publishing and Foundry governance in one walkthrough. Discuss the platforms your teams want to govern next.