Sentient-X

Main Dashboard
AI Assistant
Incident Viewer
AI Tools Registry

Inspiration

While building and selling enterprise Customer Success tools, we repeatedly encountered the same hesitation from customers.

They were willing to grant software access to CRMs and internal systems, but AI raised new concerns:

Employees using unsanctioned (“shadow”) AI tools
Prompts leaking IP or PII
No visibility into how AI was actually being used

AI adoption was already happening, but governance and visibility were missing. Sentient-X was built on the belief that AI governance must be continuous, autonomous, and enterprise-grade.

What it does

Sentient-X is an autonomous enterprise AI governance system.

It continuously monitors live AI tool usage, enriches events with enterprise context, detects risk, and responds in real time — without human prompts.

It:

Monitors AI tool usage across the organization
Classifies licensed vs. shadow AI usage
Detects potential PII and sensitive data exposure
Autonomously creates governance incidents
Explains decisions using an AI agent with live system context

The system runs as a closed loop: observe → reason → act → explain.

How we built it

We focused on real enterprise integrations and production realism.

Auth0 for employee, department, and organization lookup
AWS Lambda for telemetry ingestion and forwarding
Claude as the MCP-style reasoning agent answering governance questions using live context
Tonic Fabricate to generate synthetic telemetry logs for safe, repeatable demos
Deterministic replay via a hidden admin control plane
Fallback paths so the demo never depends on external availability

The UI reflects system state, but governance decisions are driven by the agent.

Challenges we ran into

Designing autonomy without unpredictability
Ensuring identity enrichment always mapped to real employees
Preventing synthetic data from breaking governance logic
Keeping the system explainable while acting independently

Accomplishments that we're proud of

Built a fully autonomous AI governance loop
Integrated real enterprise identity and telemetry systems
Demonstrated real-time PII and policy risk detection
Delivered a stable, deterministic, demo-safe system
Made the AI agent the decision-maker, not just an assistant

What we learned

Enterprises need visibility before governance
AI governance must run continuously, not as audits
Context (employee, org, domain) is critical to meaningful risk detection
Autonomy only works with strong guardrails and explainability

What's next for Sentient-X

Self-improving governance agents that learn what constitutes safe vs. unsafe AI usage
Policies and risk models tailored to agent behavior, not just human users
Deeper integrations with enterprise security tools (e.g., SWG)
More advanced ML-based behavioral and anomaly detection
Domain- and business-specific PII models

Our goal is simple: AI governance that runs itself, even as AI agents become first-class users inside enterprises.