Company announcement · September 8, 2026
VeriProof gives the business owner, risk and compliance, governance engineers, developers, auditors, and administrators each a portal for their part of the work, and moves exceptions, rule changes, and risk decisions between them. Every agent action it checks ends up in a record an auditor can verify on their own.
RJR Labs LLC today launched VeriProof, which helps businesses put AI agents into production and keep them there. VeriProof checks a proposed agent action against the organization's rules while the action can still be changed, sends the exceptions to the person with the authority to decide, and keeps the resulting decision record in a form an outside reviewer can verify.
Most enterprise agent programs stall between a working pilot and a production rollout. The model is rarely the reason. Work stops when nobody can say which rule applied, who approved the exception, which agents are covered, and what the agent did, so the business owner who funded the agent can't get it signed off.
"A policy document cannot control a system that takes its own actions," said Bob Janacek, founder of RJR Labs. "Neither can a log you write about yourself. What a bank or a hospital needs is the rule that applied, the person who decided, and a record their auditor can check independently. That is the product."
Most agent tooling is built for the developers shipping the agent. VeriProof gives each of the six roles that put an agent into production a portal for its part of the work: the business owner, the risk and compliance officer, the governance engineer, the developer, the auditor, and the administrator. Work moves between them through handoffs that ship today:
Each action gets one of five outcomes: Allow (continue as submitted), Warn (continue with a warning), Transform (continue with a corrected version), Escalate (wait for a person with authority), or Deny (stop). VeriProof returns the result while the connected application can still act on it, and the application stays in charge of what it does next.
The request, the rule, the review, and the outcome the application reported stay linked as one decision record. Evidence packages ship with a signed inventory and a proof reference. A reviewer verifies the signature with a trusted public key, without going through VeriProof. Each managed environment uses Azure SQL Database Ledger by default, or the Solana blockchain if chosen when the environment is created. No personal data is published to Solana.
VeriProof connects agents written with its SDKs, agents a colleague assembles in Microsoft Copilot Studio, agents working through a browser, and agents reaching enterprise systems through tools. Decisions are recorded in the vocabulary each industry already uses, drawn from published standards including ISO 20022 and BIAN in banking, MISMO in mortgage, ACORD in insurance, HL7 FHIR in healthcare, TM Forum in telecommunications, and ISA-95 in manufacturing. Agents send those field names, policies are written against the same names, and reviewers read a plain-language label with a citation to the standard behind it.
VeriProof is available today as a managed service. Private Enterprise, coming soon to Azure Marketplace, will run entirely in the customer's own Azure subscription, completely separate from veriproof.app. An Edge Gateway runs eligible policy checks beside the application for high-frequency agent workflows.
Organizations can create a free Developer account for 60 days to evaluate all enterprise software features with unlimited users and agents, 1,000 recorded decisions per month, and 14-day rolling retention. Optional paid implementation help covers agreed integration, configuration, and testing work, subject to staff availability. Production plans provide production use, higher volumes, and longer retention.
RJR Labs LLC is a bootstrapped enterprise software company based in New Jersey. VeriProof is its product for putting AI agents into production: a portal for each person who answers for the agent, policy decisions before an action, exceptions that reach a person with authority, and decision evidence that holds up to outside review.
Journalists and analysts can request a briefing or explore the platform overview.