Credit underwriting
Blocks discriminatory lending patterns, escalates low-confidence approvals, and signs every decision for regulatory audit trails.
Try this demo →Deterministic Execution Control for AI Systems
Govern every AI decision before it becomes an audit problem.
ESGF-X gives regulated AI teams runtime policy enforcement, cryptographic traceability, public verification, and operational controls from one system. The live demo, verification flow, Guardian dashboard, and pricing are live and ready to explore.
270+ tests passing
Ed25519-signed traces
Merkle-chained ledger
HMAC decision signing
Public verification route
Proof in five minutes, no signup required
Open the demo, pick a use case, watch real signed decisions stream from the live governance layer. No account, no setup, no sandboxed mock data.
<10ms
Enterprise queue p99
7 roles
RBAC in server
4 routes
Public trust surface
30 min
Integration target
Shipped customer surface
Runtime enforcement
Governance decisions happen in the execution path, not as a disconnected reporting afterthought.
Cryptographic evidence
Every trace can be signed and verified without giving an auditor direct system access.
One integration, four surfaces
Demo, verification portal, Guardian dashboard, and API docs all serve from the same deployed service — no separate tooling to stand up.
Revenue-ready billing
Self-serve pilot checkout, Stripe-metered usage, tenant provisioning, and tier enforcement are live in the same service as the governance engine.
Industry-Ready Use Cases
Watch governance solve real regulatory challenges in minutes.
The live demo shows how three different industries use the same governance layer to support compliance review without losing model performance.
Claims triage
Surfaces suspicious denials for human review, detects conflicting evidence, and maintains audit-ready decision logs for every claim.
Try this demo →Workforce screening
Prevents discriminatory hiring decisions, escalates borderline candidates into human-in-the-loop paths, and creates defensible screening records.
Try this demo →
The Evidence Argument
Replace "trust us" AI operations with evidence your buyer, auditor, and operator can actually inspect.
Unaudited AI operations create legal exposure. A single governance layer provides the runtime enforcement, cryptographic evidence, and public verification surfaces that regulated teams require.
Without governed execution
- Decision logs are fragmented across app logs, model logs, and support tickets.
- Policy enforcement is manual, inconsistent, or post-hoc.
- Audit evidence depends on screenshots, CSV exports, and human recollection.
- Sales has no live route to show what "governed AI" means in practice.
With ESGF-X
- Each decision can be intercepted, evaluated, signed, and written to a verifiable ledger.
- Risky paths can be blocked or escalated before the output lands downstream.
- Operators get dashboard, billing, and verification routes anchored to the same service.
- Prospects can move from landing page to live proof without needing a custom demo setup.
Shipped Capabilities
Capabilities built for regulated environments.
These capabilities are live, tested, and present in the deployed system — not on a roadmap.
Ed25519 signatures + Merkle chaining
Immutable trace evidence is a first-class product surface, not a hidden back-office export.
Allow, gate, escalate, or alert
Runtime decisions follow a four-outcome vocabulary: allow, gate, escalate, or log-alert. Each outcome is intentional, traceable, and visible in the live feed.
Cache, circuit breaker, HMAC
Production hardening includes TTL policy caching, circuit-breaker state machine, and HMAC-signed responses. Each contributes to measured uptime under concurrent load.
Tiering, billing, and pilot checkout
Stripe-metered billing, self-serve pilot checkout, and tenant provisioning are live. The governance engine and the revenue path are part of the same deployed service.
Dashboard + docs + verification
The Guardian dashboard surfaces tenant metrics, ledger entries, and billing status with permission-isolated views for customers, operators, and auditors.
Ready for Railway-backed production
Deploys as a FastAPI service on Railway, Docker, or bare Linux. Integration target is 30 minutes for a standard inference path. No model changes required.
Operator Workflow
The governance lifecycle in four steps.
From intercepting a model function to producing a verifiable evidence artifact — each step runs in the production request path, not as a background job.
1
Intercept
Wrap a model path with @governed so governance is part of execution, not a separate report job.
2
Decide
Route by risk and confidence, then allow, gate, escalate, or log with a concrete reason surface.
3
Sign
Persist decision evidence with cryptographic signatures, integrity checks, and verification-ready artifacts.
4
Operate
Inspect metrics in the dashboard, share proof via the verification flow, and convert pilots through billing surfaces.
30-Day Pilot
Start with one governed AI workflow, not a full platform rollout.
The pilot wraps 1-3 risky AI functions or agent actions with runtime policy checks, signed audit evidence, and a compliance-style report that supports later review.
- One tenant, one policy set, one narrow workflow scope.
- Signed evidence per governed run with dashboard and verification access.
- A clear pass/fail outcome after 30 days based on control and evidence quality.
Product Routes
Four routes. Every customer need covered.
Demo, verification, dashboard, and docs each handle a distinct part of the customer journey. All are live and accessible from the same service.
Live demo
See governance decisions stream in real time with signed trace artifacts and visible routing outcomes.
/demo Open live demo →Verification portal
Give auditors and prospects a place to verify trace evidence without giving them internal access.
/verify Open verification →Guardian dashboard
Tenant, operator, and owner workflows converge in the dashboard route with tier-aware visibility.
/app Open dashboard →API docs
Let engineering leads inspect the live schema, routes, and integration path without requesting a PDF.
/docs Read docs →
Pricing
Start with the demo. Move to a pilot when you're ready.
The demo is free and includes 50 test traces with real signed governance decisions. Pilot plans unlock testing with your own AI systems and custom policies.
Questions about fit before committing? Run the live demo first, or book a pilot-fit call. Then move to pilot checkout or a monthly plan when you're ready.
FAQ
Common questions from technical buyers.
How fast is the governance layer?
Enterprise-tier decisions complete in under 10ms (p99). This is fast enough to sit in the production model invocation path without adding noticeable latency to user-facing inference.
Can I customize the governance rules?
Yes. The policy engine is configurable, and the decision vocabulary (allow, gate, escalate, log_alert) maps to your operational needs. The demo shows the defaults; integration sessions customize these for your use case.
What proof do I get beyond logs?
Signed traces with Ed25519 signatures, Merkle-chained ledger entries, public verification routes that don't require internal access, HMAC integrity on decision responses, and verifiable evidence that governance is part of execution.
How does this integrate with my existing models?
The @governed decorator wraps your function at the boundary. No model retraining required. Integration target is 30 minutes for a single inference path. The live demo shows exactly what the code change looks like.
What happens during outages?
The system fails closed. If the governance service is unreachable, the inference path blocks instead of silently skipping governance. You can tune this behavior during implementation.
Who can see the audit logs?
Depends on your role: customers see their tenant data, operators see system-wide views, auditors get the verification portal (which doesn't require internal system access). The dashboard surfaces this with proper isolation.
What should a prospect do first?
Start at /demo, pick an industry example (credit, claims, or screening), watch the governance flow happen live. The demo is free and includes 50 test traces. After that, move to pilot checkout or choose a monthly plan to continue testing with your own data. Then review /verify for validation and /docs for integration details.
How does this support GDPR and EU AI Act review?
The system is built for Article 11 reporting and includes EU-specific classification fields in the trace schema. It supports compliance review by generating runtime evidence and policy-enforcement records that your team can inspect with counsel, auditors, or internal reviewers.
Start with the live demo. No account required.
Pick a use case, watch governance run, inspect the signed traces, then move to verification, docs, dashboard, or pilot checkout when you're ready.