Skip to main content
Verdict and EvidenceIQ

One platform.
Four steps to trustworthy authorization context.

Verdict requires human approval for high-impact agent actions. EvidenceIQ keeps the audit-ready record of every decision.

Assess where authorization lives in your code, govern policy and context together, put that context in reach of operators and agents, and defend high-risk actions. Each step stands on its own — together they move you from scattered if user.role == admin checks to decisions you can trust and evidence.

EnforceAuth is the OPA/Rego authorization control plane. The customer's OPA decides what people, services, and AI agents may do at action time. Verdict defends the high-impact action. EvidenceIQ keeps the record.

01AssessZift
02GovernWrit
03DefendVerdict
04ProveEvidenceIQ
Segmented by step

Every step is a piece of the platform you can adopt on its own.

Four stages of one platform. Here is what each one is for and where it fits in the journey.

01AssessOpen source

Zift Migration Tool

Primary use

Sift embedded authz. Extract to PaC. Rego or Cedar. Discover where authorization logic and context dependencies are buried.

Explore Zift
02Govern

Writ

Primary use

Author Rego. Ship bundles. Govern entities, drift, and audit. Improve and govern the policy and the context entities decisions depend on.

Explore Writ
03Defend

Verdict Agentic Firewall

Primary use

Decides agent tool calls over MCP, including a human approval when the action needs one. Decide agent tool calls, and require a human when the action needs one.

Explore Verdict
04Prove

EvidenceIQ

Primary use

A durable, searchable record of authorization decisions. It records them. It never makes them. Keep a durable record of those decisions. EvidenceIQ records. It does not authorize.

Explore EvidenceIQ

Decision evidence

The record an auditor can actually read

AI agent authorization evidence is the record of one access decision, kept so a person who was not in the room can check it later. A useful record answers five things: who or what asked, on whose behalf, what data or tool was involved, which policy answered, and why. A login record answers the first of those. Auditors are starting to ask for all five.

EvidenceIQ

Every time an agent asks to do something, EvidenceIQ keeps those five facts together. We can talk through the record on the contact page.

Nine frameworks, with the dates attached

Those records map to nine frameworks teams already get asked about: the EU AI Act, DORA, HIPAA, PCI DSS, NIST SP 800-53, ISO/IEC 27001:2022, SOC 2, SEC Reg S-P, and FINRA. Two dates are worth pinning down. DORA has applied to EU financial entities since January 17, 2025. The EU AI Act’s high-risk obligations under Annex III, including logging in Article 12 and human oversight in Article 14, apply from December 2, 2027.

SOC 2 is on that list as a framework a program can map to. EnforceAuth does not have a SOC 2 report. The mapping supports a compliance program. It does not replace an auditor, and it is not a certification or an endorsement.

The pressure is not theoretical. In Gravitee’s April 2026 survey, 54% of organizations said they had experienced or suspected an AI agent security or data privacy incident in the past 12 months.

The longer version, for the person who has to answer the auditor.

Platform FAQ

How the suite fits together

Verdict requires human approval for high-impact agent actions. EvidenceIQ keeps a durable, searchable record of every authorization decision. Identity can be right, and the rule can be right, and the decision can still be wrong if ownership, classification, or delegation was missing or out of date. That shows up first with AI agent authorization, and it was already there for apps, APIs, and human access.

AI agent authorization evidence is a record of one access decision that someone else can check later. A useful record says who or what asked, on whose behalf, what data or tool was involved, which policy answered, and why. A login record only has the first of those. EvidenceIQ keeps that record. It records decisions and does not make or change them.

Assess with Zift, Govern with Writ, Defend with Verdict, and Prove with EvidenceIQ. Each step can stand alone. Together they are one path to governed, evidence-backed authorization on OPA. EvidenceIQ keeps the record.

Most teams start with Zift to find embedded authorization logic, carry generated Rego into Writ, then add Verdict and EvidenceIQ once policy and the context it depends on are governed together.

No. EnforceAuth authorizes humans, services, and AI agents. AI agents are where the context problem becomes obvious — they are not the boundary of the category. The same discipline applies to applications, APIs, Kubernetes, and human access. We started from authorization context, not from an AI-only product.

No. OPA evaluates policy against supplied context. EnforceAuth helps you trust that context — and govern authoring, bundles, entities, and agent tooling around the OPA fleet you already run.

Start anywhere. The path connects.

Most teams start with Zift to see sprawl and authorization context debt in code, then carry generated Rego into Writ. Verdict and EvidenceIQ layer on once policy and the context it depends on are governed together.