00
Days
:
00
Hrs
:
00
Min
:
00
Sec
EU AI Act
Enforcement deadline

Your AI agents are deploying.Who governs them at runtime?

PhronEdge enforces governance at the agent boundary. Every tool call passes through 7 constitutional checkpoints before execution. Every decision is hash-chained and cryptographically provable.

A policy document cannot block a tool call. A maturity model cannot redact PII from an output. A framework slide deck cannot revoke a credential when an agent is compromised. PhronEdge fires at runtime. That is the difference.

Live Governance

Seven checkpoints. Every tool call. No exceptions.

This agent tried to access patient medical records without clearance. The pipeline blocked it at checkpoint 6 and cited GDPR Art. 9.

PhronEdge governance pipeline blocking unauthorized access
7 checkpoints on every call
Credential validation, jurisdiction resolution, PII detection, behavioral analysis, tool permissions, data classification, output constraints. Every request. Every time.
Immutable audit chain
SHA-256 hash chain. Every governance decision anchored with cryptographic proof. Tamper one event and the chain breaks. Your auditor trusts the math.
Real-time agent lifecycle
Quarantine suspends all tool access. Kill switch revokes the credential globally. Tamper detection catches credential modification and auto-restores from vault.
Any model. Same governance.
OpenAI GPT-4o, Google Gemini 2.5, or your own. The governance layer is model-agnostic. Switch providers mid-session. Same policy. Same enforcement.
Launch Playground

Insurance, Healthcare, Finance, Technology. No signup required.

Constitutional Console

Build policies. Watch enforcement. Export proof.

ObserverPolicyAudit LogArchitecture
claims-investigator
active
342
calls today
medical-reviewer
active
128
calls today
fraud-detector
quarantined
0
calls today
ALLOWEDsearch_claimsclaims-investigator
ALLOWEDlookup_patientmedical-reviewer
BLOCKEDaccess_recordsclaims-investigator
Developer Tools

Three lines. Full governance.

pip install phronedge. Govern your tools. Scan for ungoverned functions. Verify your connection. Three commands, full coverage.

your_agent.py
$ pip install phronedge

from phronedge import PhronEdge
pe = PhronEdge()

@pe.govern("lookup_claim")
def lookup_claim(claim_id: str) -> str:
return db.query(claim_id) # governed. 7 checkpoints. <50ms.
LangChainCrewAIGoogle ADKOpenAIPydantic AIAutoGen
36+
Regulatory frameworks
196
Jurisdictions mapped
7
Runtime checkpoints
<50ms
Gateway latency
ECDSA
Cryptographic signing

One policy. One signature. Full governance.

You declare what you have. PhronEdge evaluates, signs, and enforces.

Declare
Define your agents, tools, and permissions. PhronEdge maps your declarations to 36+ frameworks automatically.
Sign
ECDSA P-256 signature anchored on-chain. The signed credential is immutable. We delete the policy after signing. You own it.
Enforce
Your agents carry their signed credential. The 7-checkpoint gateway validates every action before execution.
Prove
Every decision is SHA-256 hashed and chained. Your regulator gets a cryptographic proof chain. Not a report.

Five components. No single point of compromise.

Separation of powers applied to AI governance.

The Brain
Evaluates policy. Signs with ECDSA P-256.
→
The Gateway
7 checkpoints. Every call. Under 50ms.
→
The Anchor
SHA-256 hash chain. Tamper-evident.
→
The Enforcer
Kill switch. Quarantine. Credential revocation.
→
The Observer
Real-time telemetry. Anomaly detection.

Built for regulated industries.

Each sector gets a tailored governance overlay traced to its specific regulatory clauses.

Financial Services
DORA, SR 11-7, BCBS 239, MiFID II, OCC, FFIEC
Healthcare
HIPAA, HITECH, FDA AI/ML, 21 CFR Part 11
Insurance
NAIC, Solvency II, EIOPA, State AI Regulations
Defense and Aerospace
FedRAMP, CMMC, ITAR, DFARS
Pharmaceutical
FDA 21 CFR Part 11, GxP, EMA Guidelines
Enterprise Technology
ISO 42001, SOC 2, FedRAMP, EU AI Act, GDPR
NIST AI RMFISO 42001SOC 2 Type IIEU AI ActOWASP LLM Top 10OWASP AgenticDORAGDPRHIPAAFedRAMP

Your policy. Our signature. Stored in your vault. Proven on-chain. We never hold it after signing. If we go down, your signed policy still runs.

Your agents are running.
Are they governed?

Try the playground. Read the docs. Deploy in production.