Bob Janacek
Bob Janacek founded RJR Labs and is building VeriProof, its first venture. He has spent about thirty years building regulated software, working in enterprise security, and leading companies. RJR Labs is bootstrapped.
About VeriProof
Built to get AI agents into production and keep them there
RJR Labs builds VeriProof for the business owner who funded an AI agent and needs it doing real work: approved, trusted with more over time, and able to answer for its decisions when someone asks.
Why we exist
There are plenty of good tools for the developers building AI agents, and they do that job well. They produce very little for the other people an agent depends on: the business owner accountable for the results, the compliance officer who has to approve it, the reviewer who handles its exceptions, and the auditor who arrives later.
So agents that work in a demo wait for a sign-off nobody can give. VeriProof gives each of those people a portal for their part of the work, and moves the work between them. Each agent action is checked against policy while the application can still act on it, sent to a person with authority when it should be, and recorded in the vocabulary of its industry, in a form an outside reviewer can verify.
The result we're after is agentic AI that gets out of pilot, into production, and stays there.
Founder experience behind VeriProof
Secure enterprise software at scale
Bob spent 27 years building and leading DataMotion, including as founder and CTO and later CEO. He previously founded Safetynet, where he developed security and network automation systems for enterprise and U.S. military customers. He also serves on the Information Systems & Analytics advisory board at Stevens Institute of Technology.
How we build
A few choices run through the whole product. Each one came from watching regulated software fail an audit, or pass one.
Rules are checked before the action runs
The policy check happens while the application can still change course, and every check creates its record automatically. Nobody has to remember to log it.
Your application stays in charge
VeriProof returns one of five outcomes: Allow, Warn, Transform (continue with a corrected version), Escalate (wait for a person with authority), or Deny. Your application applies it and reports what it did, so the record holds both the decision and the outcome.
Rule changes on the record
A rule has to pass its tests before it's enforced, and Shadow can show what it would decide without changing what the agent does. You choose how many people besides the author approve a change. Production changes from governance engineers and developers always go to an administrator or compliance officer.
Exceptions go to authority, not availability
An exception reaches the person who holds the authority for it, in the VeriProof Inbox, by email, or in Slack or Teams, with reminders and escalation. Routine calls can be decided from the Slack message or Teams card itself, and the signed answer returns to the work that was waiting.
Evidence a reviewer can check
A reviewer outside your team can check a record without taking our word for it. Here is how. Each environment uses Azure SQL Database Ledger by default, or Solana if you choose it when the environment is created. No personal or patient data is ever published to Solana.
Code you can read if you need to
The SDKs and platform use the standard VeriProof software license, and source is available on request. SaaS plans get the security review tier, to read and inspect the code. The modification tier, to change it and run it in your own environment, is part of Private Enterprise, coming soon to Azure Marketplace.
Bring one agent you want in production
Start a free 60-day Developer evaluation with your work email, or look at a week of one agent's work first. Developer includes unlimited users and agents, within its recorded-decision and retention limits.