Governance
The same event stream that blocks an action also proves your controls
Security teams govern agent controls with the same rigor as infrastructure code, while audit and finance read continuous evidence from the identical records — no duplicate integration work.
Policy-as-code
Policy-as-Code
Enterprises need policies that can be reviewed, versioned, tested and promoted through environments. Policy-as-code complements the visual builder and makes controls fit modern engineering and security change-management processes.
What we build
- Human-readable DSL/YAML or declarative policy representation.
- Git-backed version control and pull-request workflow.
- Policy unit tests and fixture library.
- Environment promotion: dev → test → production.
- Simulation against historical events before release.
Implementation decisions
- One canonical policy compiler for both UI and code-authored rules.
- Validate syntax, references and dangerous precedence changes in CI.
- Support signed policy bundles and gateway verification.
- Provide reusable policy packages/templates without hiding generated code.
ENTERPRISE FIT
Security teams should be able to govern agent controls with the same rigor as infrastructure and application code.
Control mapping
Compliance & Governance Mapping
Agent Trust Cloud should help customers operationalize AI governance frameworks without presenting itself as legal advice or automatic compliance certification. The product can map controls and evidence to recognized frameworks and regulations, while customers remain responsible for applicability and legal interpretation.
Frameworks we map to
- Control mapping library for NIST AI RMF, ISO/IEC 42001, NIST zero trust concepts and common security-control frameworks.
- EU AI Act workflow support for inventory, documentation, risk records and incident/evidence processes where relevant.
- Customer-specific internal AI policy mappings.
- Gap dashboard: required control, implemented technical control, owner, evidence and exception.
How we keep it honest
- Version framework mappings and show source/version dates.
- Separate “technical evidence available” from “compliant” conclusions.
- Allow customers/partners to add legal interpretation notes and applicability decisions.
- Build exportable control/evidence packages for audit workflows.
CURRENT BASELINE
NIST AI RMF supports AI risk management; ISO/IEC 42001 defines an AI management system; EU AI Act obligations continue to phase into enforcement in 2026.
Audit automation
Audit Reporting & Evidence Automation
Audit value comes from reducing manual evidence collection. The platform should turn runtime records, approvals, registry state and policy versions into repeatable evidence packages for internal audit, external audit and governance reviews.
Reports and exports
- Scheduled access/agent inventory reports.
- Privileged-agent control coverage report.
- Policy exception and approval report.
- Agent lifecycle/recertification report.
- Incident and containment report.
- Model/provider usage and data-policy report.
Definition of done
- Auditor can sample an action and trace it to source evidence.
- Recurring report can be generated without manual log collection.
- Report generation respects tenant retention and data-minimization settings.
COMMERCIAL VALUE
Compliance reporting turns the same security event stream into recurring executive and audit value.
Also in this area
Operations, analytics and FinOps governance
Observability & Telemetry
Observability must capture what agents attempted, why controls responded, and what ultimately happened. The telemetry system should serve security operations without becoming an uncontrolled warehouse of sensitive prompts and business data.
OPERATING PRINCIPLE →
Agent Activity Recorder & Replay
The activity recorder is the black box for agent behavior. It reconstructs sequence, context and authority so an incident responder or auditor can understand how a result occurred across multiple tools and agents.
FORENSICS →
Behavioral Anomaly Detection
Behavioral detection identifies suspicious deviations that static policy cannot anticipate. The design should combine understandable statistical baselines with optional machine-learning models and never let anomaly scoring alone execute irreversible remediation without policy.
DETECTION RULE →
Agent Risk Scoring
Risk scoring gives leaders a consistent way to prioritize remediation. The score should reflect inherent capability, data exposure, autonomy, identity assurance, policy coverage, behavior and control maturity—not a mysterious single AI prediction.
EXECUTIVE VIEW →
Incident Response
Agent incidents can span identity, model, tool, data and business systems. Agent Trust Cloud should package the relevant context into a case workflow and automate containment actions while keeping irreversible decisions under explicit policy.
SOC VALUE →
Kill Switch, Quarantine & Containment
Customers need confidence that they can stop an agent immediately. The kill switch must be fast, scoped and resilient, with options ranging from one session to an entire agent class or tool integration.
PROMISE →
Forensics & Evidence Chain
For high-stakes investigations, evidence needs integrity, chronology and traceability. The platform should maintain a tamper-evident chain that shows what was observed, what policy was active, who approved actions and what downstream system returned.
TRUST →
AI FinOps & Spend Governance
Once Agent Trust Cloud observes model and tool usage, it can give organizations a unified view of AI spend by agent and business purpose. This creates a second economic buyer and turns security telemetry into operational savings.
EXPANSION MOTION →
Budgets, Quotas & Model Routing
Budget enforcement should control not only dollars but also rate, concurrency and model selection. Policy-aware routing can reduce cost while preserving required quality, security and residency.
FINOPS CONTROL →