Santry · AI Agent Governance
Clear identity. Defined permissions. Traceable actions.Let agents collaborate and act within the boundaries your enterprise sets.
Why governance matters
A task can pass between agents. The original requester and the responsible team need to stay identifiable.
Access to a system does not authorize every action. A specific payment or data change needs its own authorization.
When requests, approvals and system results live in separate places, reconstructing an operation becomes difficult.
How Santry works
Santry gives every agent a distinct identity and connects access permissions, action authorization, and activity records, so your organization can see who is acting, what they are authorized to do, and the outcome.
01 / Get a badge
Santry registers each agent, issues a dedicated identity credential, and records its team, responsibilities, and accountable owner. By linking to your existing enterprise identity system, Santry gives every agent a clear identity and an accountable owner.
Identify both the person making the request and the agent carrying it out.
02 / Check access
Santry checks an agent's access to tools, systems, and environments against your organization's rules. Only authorized resources are accessible; unauthorized requests are blocked by default, keeping collaboration within clear boundaries.
When a task passes between agents, each agent must independently pass permission checks, with the original request's source preserved. Collaboration can continue without expanding permissions.
03 / Act within authorization
Santry identifies critical actions according to your organization's rules and routes requests to the responsible person for approval. Once approved, Santry checks that the authorization is valid and matches the action's details and scope. Only then can the agent proceed.
The badge represents identity and standing access permissions. Action authorization is limited to the particular operation and target approved this time.
04 / Review the record
Santry connects each action's request source, authorization basis, and execution result. It records who initiated the request, who approved it, which agent carried it out, and the outcome, providing evidence for ongoing oversight, compliance reviews, and accountability.
Follow the original request through to the returned result, with evidence for investigation and business review.
Santry in practice
Apply the story's identity, access, action authorization, and records to this HK$12,800 payment. Simulate an account change to see where Santry stops the action.
A procurement colleague requests payment. The finance assistant prepares the application, and a separate payment agent handles execution.
From the story: 01 · Receive a badgePayment request PAY-1048
Identity linkedRequester
Procurement colleague
Finance assistant
Order Agent
Payment execution
Payment Agent
Agent owner teamFinance team
Handover keeps the original request linked. The receiving agent's identity and permissions are checked again.
What if the account changes?
Keep the approval for 4821, but change the instruction to 9067.
A business demonstration of the workflow, not the actual product interface or a customer case. Records shown cover operations governed through Santry.
Working with your systems
Start with the agents, tools, and workflows your team already uses.
Connections at a glance
Your agents
Calls & handoffs
Santry
Permission checks
Activity records
Tools & systems
MCP tools
Business services
Your enterprise processes
Enterprise identity
Approval workflows
Connect existing agents, MCP tools, and business services. Santry checks permissions for the calls and task handoffs that pass through it.
Keep your enterprise identities and approval workflows. Business systems continue to enforce their own permission checks.
Deploy to suit your environment
Deploy and operate Santry in your own environment.
Use the hosted service and reduce infrastructure maintenance.
How Console, Registry, Policy, and AgentGateway support the governed request path.
Request origin
Original intent + caller context
Santry governance
Console
Configure agents and governance
Registry
Find agents & services
Policy
Evaluate access & task scope
AgentGateway
Verify · decide · route
Connected environment
Only within the authorized scope
A connected audit trail
Original request · participating agents · authorization decision · execution context
Choose a workflow and verify that permitted actions proceed, unauthorized actions are stopped, and approval records can be reviewed. Then extend the proven rules and connections to more teams.
Discuss a Santry demoBring one workflow, the agents and tools involved, and your current permission and approval rules. Together, we can identify the actions that should proceed, the ones that should be blocked, and the records needed for review.
No. Routine tasks can run automatically under your enterprise rules. Sensitive actions, such as payments or data changes, follow your existing approval process.
