What to detect
Credential patterns, personal information and enterprise-defined fields, according to supported rules and versions.
Product /Route Fence
Define what AI can see and do.
Place sensitive-data checks, model access controls and tool authorization in the actual execution path to enforce enterprise rules.
Built for enterprise
01
Request side
Check documents and context before they are sent.
02
Execution side
Check tool actions, target resources and permissions.
03
Traceability
Record policy decisions, approvals and event outcomes.
01 / Risk across the path
Check where data and actions pass through.
02 / Context protection
Handle sensitive content before it leaves.
Identify enterprise-defined sensitive fields, then decide handling based on the destination model and task scope.
Handling options
Block, redact or allow; exceptions follow an explicit authorization process.
Destination restrictions
Check allowed models, addresses and resource scopes instead of sending to arbitrary endpoints by default.
Business impact
Redaction may affect quality. Validate false positives, missed detections and usability.
03 / Tool-execution protection
Let access rules decide whether a proposed action runs.
Check identity, parameters and target systems before execution, with human approval for high-risk actions.
Allowed scope
Tool allowlists, target domains, commands and parameter constraints.
Least privilege
Provide only the permissions needed for the current task.
Credential use
Provide credentials through controlled mechanisms; short-lived or task-bound support depends on the version.
Human approval
Send sensitive configuration changes and high-risk actions to the responsible owner.
04 / Enterprise-side data plane
Make processing location part of the architecture.
Configure gateways, credentials and logs to enterprise needs, with clear connections to management services and external models.
Enterprise gateway
Plan controlled request processing in an enterprise intranet or VPC.
Control services
Define policy distribution, metadata reporting and operational access.
Your own credentials
Use enterprise-authorized model keys, with explicit access and storage arrangements.
Log retention
Retain metadata and policy outcomes as needed; full-content logging is configured separately.
05 / Audit and incident response
Build reviewable event records from policy decisions.
Link requests, tool actions and approvals to investigate issues and refine rules.
Event fields
Time, user or application, agent, model, tool, policy outcome and correlation ID.
Response process
Detect → block or request review → alert → trace → review rules.
Access scope
Control log access through permissions and mask sensitive content where needed.
What records can establish
Audit supports investigation; it neither replaces all real-time blocking nor proves a path is entirely tamper-free.
06 / Coverage and next steps
Define the scope before designing the controls.
Map model entry points, agents and business tools, then locate policy enforcement for each call type.
Integrated requests
Model calls through the enterprise gateway can be checked according to configuration.
Integrated actions
Tool actions integrated at the execution endpoint can enforce parameters and permissions.
Unintegrated paths
Direct model calls and unintegrated tools need separate inclusion; they are not covered by default.