Phinn

every agent action, accounted for

The enforcement layer for AI agents

Phinn sits between your agents and tools (HTTP + MCP), enforcing default-deny, requiring human approvals for risky actions, and generating cryptographically verifiable receipts for every call.

AI agents now hit production without real controls

Four ways it breaks down.

Unmanaged access

LangGraph, Claude, and ChatGPT can hit APIs, databases, and internal services with no enforceable guardrails.

No audit trail

When an agent mutates production data, teams can’t prove who initiated or approved it.

Bypass risk

Agents can call tools directly unless enforcement sits at the gateway layer.

Zero visibility

SecOps lacks real-time, tamper-evident telemetry on agent behavior.

Enterprise-grade enforcement for AI agents

01

Default-deny enforcement

Phinn issues short-lived, capability-scoped tokens per action. No token → no execution.

02

Human approvals where work happens

Slack approvals for deletes, wires, and other risky actions. Policies define approvers. Every decision is recorded with identity.

03

Tamper-evident receipts

Hash-chained, Ed25519-signed receipts for every action. Verify offline with phinnctl verify. Prove to auditors exactly what happened, when, and who approved it.

04

SIEM-ready telemetry

Structured events stream to Splunk or Datadog using OTel semantics for full agent visibility.

Built for Security, Platform, and Compliance teams

Phinn runs entirely in your infrastructure. Tool traffic stays in your VPC; only configured telemetry and approvals leave. Fully self-hosted control plane today, managed issuer on deck.

How it works

Three components working together: gateway, issuer, and your existing tools.

Agent
LangGraph, Claude, etc.
Phinn Gateway
Default-deny enforcement
Human Approval
Slack / Issuer
HTTP + MCP Tools
DB, APIs, etc.

1. Agent requests capability

Agent hits the issuer for a token scoped to method, path, and tool.

2. Policy & approvals

Issuer runs policy and, if needed, triggers Slack. Token is minted only after approval.

3. Gateway enforces

Gateway verifies binding, prevents replay, forwards to the tool, and writes a signed receipt.

Deploy anywhere

Same gateway, multiple topologies.

Kubernetes sidecar

Run the gateway beside each agent workload. Route HTTP/MCP calls with minimal config changes.

Cluster gateway

Operate Phinn as a central ingress so agent traffic routes through one enforcement point.

VM in your VPC

Drop it on a VM for non-Kubernetes environments. Same enforcement, simpler stack.

Contact Us