Menu

Product / AgentGuard

Guard your agent. Prove what it did.

See the whole run. Apply the right controls. Keep the evidence.

ONE REQUEST / THE WHOLE RUN

Reason freely. Act within bounds.

THE REQUEST“Prepare a payment to a new payee.”

TRACEPROTECTCONTROLEVALUATEPROVE
Banking agentCustomer request
Customer contextSensitive fieldsCHECKING
Policy contextTool permissions
ModelPlan the action
Account checksRead-only toolPOLICY CHECK
Payment toolAction heldPOLICY CHECK
ResponsePayment not sent
EvaluationOutput checkedCHECKING
Request received
TRACE

Banking agent

Customer request

Decision stays with the run

The customer request opens one trace.

Illustrative runs using configured controls. Actions remain held; no payment, care or operational change is executed.

Set the controls.
Guard the run.

The AgentGuard Controls screen brings model access, tool permissions and resource limits into one place.

AgentGuard Controls screen showing model access, tool permissions, token limits, rate limits, budgets and execution limits
Actual product capture · Controls / Agent PoliciesExplore AgentGuard ↗

FITS THE WAY YOU BUILD

Your stack.
Your deployment.

Built for multi-model, multi-framework AI environments.
Deploy your way.

Models

OpenAI

Anthropic

Verified SDK integrations

Frameworks & protocols

LangChain

LangGraph

MCP

Verified SDK integrations

Deployment

SaaSBYOCVM

Compliance Packs

Runtime evidence for
the frameworks that matter.

AgentGuard automatically collects and maps live runtime signals — guardrail decisions, evaluations, model inventory, and audit activity — to AI-relevant controls. Start with core frameworks, then add domain packs for regulated environments.

Core frameworks

Enabled by default for new projects
  • EU AI Act

    Risk, logging, oversight, and evidence for high-impact AI use.

  • NIST AI RMF

    Map runtime activity to govern, measure, and manage functions.

  • ISO/IEC 42001

    Support AI management system evidence with runtime-backed signals.

  • India DPDP

    Map personal-data handling controls and evidence to runtime activity.

  • OWASP LLM Top 10 (2025)

    Track runtime protections relevant to common LLM application risks.

Domain packs

AI-relevant controls.
Sector-specific context.

Aviation
FAA 14 CFR Part 5
Healthcare & Life Sciences
HIPAA Security Rule, ONC HTI-1
Financial Services
SR 11-7, OCC 2011-12, FINRA RN 24-09, SEC Rule 17a-4, GLBA, NYDFS Part 500, ECOA / FHA
Insurance
NAIC Model Bulletin
EU High-Risk — Deployer
Coming soon

Coverage reflects observed runtime evidence, not a certification or legal determination.

From the first model call to the final action, AgentGuard keeps the run, the controls and the evidence connected.

Start with AgentGuard ↗