A one-time audit catches what you had on the day of the test. The AI-Agent Guardrail Retainer keeps watching — quarterly re-tests, prompt-injection regression checks, and live CVE monitoring for your agent stack, every month.
Book a retainer scanCustomer-facing agents change constantly — new tools, new prompts, new integrations. Each change re-opens the attack surface. A retainer keeps the assessment continuous so a fix stays fixed.
Full re-run of the prioritized attack paths against your live agent every quarter, not just at onboarding.
Every release gets replayed against the injection library that broke it before — so a patched hole stays patched.
We track disclosed vulnerabilities and behavior changes in the models, frameworks, and tools your agent depends on.
One maintained document your security and compliance teams can point auditors at — always current, never stale.
All tiers include the living risk register and alerting. Scale up as your agent footprint grows.
Entry point. We scope one agent and run three prioritized attack paths with written remediation guidance.
Findings become the first living risk register and the regression set for future checks.
Quarterly re-tests and CVE watch keep the register current and the holes closed.
When scope grows, the $5K full audit and ISO 42001 gap assessment slot in alongside the retainer.
The retainer builds on a baseline. Book the Founding AI Agent Risk Review and we'll scope the retainer from your first findings.
Book the $750 Review