Behavioral charter engine
Define what each agent is allowed to do, spend, and access. Charters are the source of truth — every detection and response references them.
AegisQ AI Sentinel
Universal behavioral monitoring for software AI agents and physical AI systems. Behavioral charters, kinematic safety charters, real-time enforcement (L1–L5), sensor attestation, fleet coordination, and quantum-resistant compliance proof. One platform for both surfaces.
Software agents on LangChain, CrewAI, OpenAI, and Anthropic make autonomous decisions without guardrails or audit trails. Collaborative robots, AMRs, and drones operate in shared human spaces with kinematic safety risks software alone can't govern. When connectivity drops, cloud-based governance disappears. AI Sentinel is one platform for both surfaces — software and physical AI — under one compliance framework, with cryptographic, quantum-resistant proof.
Six core capabilities define what each agent is allowed to do, detect when it drifts, and respond proportionally — every action logged and signed.
Define what each agent is allowed to do, spend, and access. Charters are the source of truth — every detection and response references them.
Behavioral profiling, capability drift, self-modification monitoring, intent inference, and digital-twin comparison — multi-layered, in real time.
From passive observation to emergency halt. Five severity levels with full audit trails — proportional response for software agents and robots alike.
Liveness probes, canary rules, and alert-flood detection ensure monitored systems can't tamper with or evade Sentinel's monitoring boundary.
ML-DSA-65 (FIPS 204) signed compliance reports — tamper-evident, quantum-resistant evidence that every system operated within charter.
DID-based identity, W3C Verifiable Credentials, and trust scoring with reputation decay — a verifiable reputation system per agent.
Cost attribution and budget enforcement per agent, granular spending controls, financial guardrails.
Full delegation graphs, authorization chains, and regulatory export for audit compliance.
Automated test suites, risk scoring, and actuarial data for AI-insurance readiness.
Track agent spawning, permission inheritance, and parent-child relationships across runtimes.
As of v3.9.0, AI Sentinel governs the Model Context Protocol surface that modern agents depend on — the tools they call and the telemetry they emit — with all phases of the OWASP Data Security for GenAI (DSGAI) 2026 framework implemented.
A permission-scoped MCP event sink governs which agents and tools may report telemetry. Charter violations resolve into typed INTERVENE / HALT action records that emitters retrieve and enforce per tenant.
MCP detection rules are backed by live OSV.dev advisory data, flagging known-vulnerable MCP server and tool versions across the environments AI Sentinel governs.
All phases of the OWASP Data Security for GenAI framework are implemented — runtime content inspection and GenAI data-loss controls across the telemetry pipeline.
Six robot families. Six governance capabilities. One control plane. Built for environments where safety boundaries are physical, not just behavioral.
UR, Franka, KUKA iiwa cobots operating alongside humans.
KUKA KR, ABB IRB, Fanuc — high-payload production robots.
Autonomous mobile robots for warehouse, logistics, and facility navigation.
Bipedal humanoids in manufacturing, service, and research.
Precision medical and surgical systems in clinical settings.
Aerial drones for inspection, delivery, surveillance, and mapping.
Joint limits, velocity caps, force thresholds, proximity zones, and geofence boundaries — enforced in real time with pre-execution validation.
Merkle-proof integrity, drift detection, and tamper alerts for every sensor on every robot. Cryptographic proof of sensor health at measurement time.
URDF-based robot models for pre-execution validation, collision prediction, and what-if simulation before motion commands hit hardware.
Shared workspace monitoring, emergent risk detection across robot types, fleet-wide emergency stop. Govern multi-robot environments as a unified system.
Robots continue safe operation during connectivity loss with time-limited cryptographic leases. When leases expire, hardware failsafes engage automatically.
Real-time safety event logging, incident classification, and automatic regulatory reporting. 7-year retention for OSHA and ISO 10218 compliance.
Cloud-based evaluation, real-time charter updates, fleet coordination, and streaming telemetry. Optimal governance with complete visibility.
On-robot agent with cached charters and store-and-forward telemetry. Safety enforcement continues locally when connectivity is limited or intermittent.
Cryptographic lease system with hardware failsafe integration. Local enforcement, zero network dependency, leases expire to fail-safe state.
Every response action is logged with full audit trail. The same five-level scale applies to software agents and physical robots.
Continuous observation with zero intervention. All activity and telemetry logged but no restrictions applied. Baseline profiling for agents and robots.
Software AI + Physical AIReal-time notifications on anomalies and charter violations. Teams receive alerts for review and manual intervention without system lockdown.
Software AI + Physical AIReduce agent rate limits or restrict robot velocity / force envelopes. Systems remain operational within tighter safety boundaries.
Software AI: rate limits · Physical AI: velocity / force capsPause agent execution or command robot to safe-hold position. Systems suspended pending human review and remediation.
Software AI: suspend execution · Physical AI: safe-hold positionEmergency shutdown of the agent and all child processes, or fleet-wide emergency stop for robots. Reserved for critical safety breaches.
Software AI: terminate + children · Physical AI: fleet e-stopAegisQ AI Sentinel is engineered design-aligned with the regulatory frameworks below. SOC 2 Type II audit is targeted for H1 2027 and ISO 27001 certification will follow. Until those certifications are complete, framework alignment is by design and self-attestation only.
ML-DSA-65 (FIPS 204) JWT signing, ML-KEM hybrid TLS for pod-to-pod communication. Future-proof against CRQCs.
From shared row-level security to fully isolated clusters. Choose the level that matches your security and regulatory obligations.
Protected innovations across behavioral governance, kinematic safety, quantum-resistant attestation, and multi-robot coordination.
A base platform fee with included agents, then volume-discounted agent blocks as you scale. Physical AI monitoring is an add-on from the Professional tier up.
$0
2 agents · 5,000 events/day · 1 tenant · Tier 1 shared isolation · 7-day audit retention · community support. For evaluation and development.
Get started$149 / mo
10 agents · 100K events/day · Email + Slack alerts · Tier 1 isolation · 30-day retention · email support (48hr). For startups and small teams.
Start a trial$599 / mo
50 agents · 1M events/day · all detection + trust scores · Tier 2 per-tenant keys · OIDC SSO · 90-day retention · Physical AI add-ons · email + Slack (8hr).
Start a trial$1,799 / mo
200 agents · 5M events/day · Tier 3 schema isolation · SAML SSO · behavioral profiling · FinOps + certification API · 365-day retention.
Talk to salesFrom $5,000 / mo
500+ agents · unlimited events · Tier 4 dedicated database · custom detection rules · 99.9% SLA · 1–7yr retention · 24/7 dedicated support.
Talk to salesFrom $10,000 / mo
Everything in Enterprise · Tier 5 cluster isolation · CMEK · FedRAMP High / ITAR / SOC 2 Type II · WORM audit · self-hosted or dedicated GCP project.
Talk to sales
Additional agents from $0.50–$2.50/agent in volume-discounted blocks. Annual plans save ~17% (2 months free).
Physical AI add-ons (Professional+): Collaborative arm $299/mo · Industrial arm $499/mo · AMR $199/mo · Humanoid $999/mo · Surgical $1,499/mo · Drone $149/mo
30 minutes to walk through behavioral charters, kinematic safety, and L1–L5 response on your own agents and robots.