What is VeriProof?

VeriProof connects AI application activity with policy checks, human review, and evidence. Teams use these records to investigate exceptions, make authorized decisions, and prepare evidence for a defined scope.

VeriProof works alongside your application and model providers. It does not host the model or require every model request to pass through a VeriProof gateway.

What the platform records

What VeriProof records depends on the SDK and adapter you use, what your application sends, and your governance settings.

Record	Examples
Application context	Application, environment, service, and correlation identifiers
AI activity	Model calls, tools, workflow steps, and agent handoffs captured as spans
Governance context	Policy results, risk fields, oversight signals, and review requirements
Decisions and outcomes	Recorded choices, authorized reviews, callbacks, and outcome data
Evidence integrity	Export manifests, commitments, and proof-provider material

VeriProof can evaluate only the evidence your integration sends. Confirm coverage with representative sessions before relying on a dashboard, rule, or export.

Where VeriProof fits


Your AI application
        |
        | OpenTelemetry spans and governance attributes
        v
VeriProof SDK or adapter
        |
        v
Ingest, policy, review, and evidence services
        |
        +--> Customer Portal
        +--> Evidence exports
        +--> Configured proof provider

The SDK exports records outside the model request path. Governance features that protect an action, such as human review, use their documented decision and callback workflow.

What teams use it for

Review the captured steps, decisions, policy signals, and available evidence for a session.

Investigate AI activity

Assign work, review exceptions, track remediation, and record authorized decisions.

Manage governance work

Export records for a defined application, period, purpose, and reviewer.

Prepare evidence

Check an export manifest and its configured provider proofs independently.

Verify integrity

What VeriProof does not decide

VeriProof does not determine whether an AI system is lawful, fair, clinically appropriate, compliant, or ready for production. Those decisions remain with your authorized owners and reviewers.

The records support those decisions. They do not replace your policies, risk analysis, legal advice, testing, or independent audit work.

Start with a development application

Create or select a development application, install the approved SDK package, and send representative non-production sessions. Then confirm the expected records in the Customer Portal.