Human services — AI agent audits
We don't just build safe agents. We audit yours.
Independent security and cost audits for any AI chatbot, agent, or automation — even ones not built on Aether. 48-hour turnaround. We bring the sandbox: your team sets up nothing, production is never touched, and we never ask for write access to your systems.
Air Canada held liable for its chatbot's hallucinations (2024 tribunal ruling)
An AI agent deleted a company's production database in 9 seconds (PocketOS, 2026)
Agentic workflows burn 5–30x more tokens per task than chatbots (Gartner, 2026)
How it works
1Tell us which agents you run
2We test them in our sandbox, never production
3Get ranked findings and fixes in 48 hours
We built the nine-layer Employee Stack so our agents are safe by design. Now our specialists turn that same rigor on yours.
Starting prices
Snapshot
$2,500
one-time
1 agent, 48-hour turnaround
- Core red-team pass
- Cost-waste scan
- Severity-ranked findings report with fixes
Deep
$6,500
one-time
Up to 3 agents
- Full test battery (red team + stress + data & permission tests)
- Sandbox testing we set up
- Compliance mapping (EU AI Act, LGPD/GDPR)
- 60-minute remediation walkthrough
Full
$12,000
one-time, first month
Up to 10 agents
- Everything in Deep
- Permanent sandbox environment
- Monitoring and spend alerts configured
- Team training
- 90-day hardening roadmap
Sentinel
$2,500
/month retainer, 3-month minimum
Available after a Deep or Full audit
- Regression testing after every agent update
- Monthly adversarial re-tests
- Managed spend monitoring
- Quarterly deep audit
- Priority incident response