AI Governance as Code: From PDF Policies to Pipelines

AI governance written into a PDF protects nobody. Ken Johnston, VP of AI and Data at Envorso, and Bob Rapp, Principal AI Architect at the General Motors AI Center, co-founded the AI GovOps Foundation to make the case that governance has to work like engineering: version controlled, automated, testable, and enforced inside the deployment pipeline rather than reviewed alongside it. They explain what governance as code changes in practice, which controls can be automated first, how to test systems that behave differently on every transaction, and why treating governance as paperwork guarantees it gets skipped under deadline pressure.

Key takeaways

•   Policies belong in the software repository next to the models. If every change does not pass the same automated checks for lineage, fairness, and risk, you are not shipping policy as code

•    Every gate has a yes or no answer, a contract, and a test. Fail the test and you do not pass the gate

•    CI/CD gates are not enough for agentic systems. Governance watchdogs also have to run at runtime, at the edges between agents, and in front of prompt injection

•    Governance done right is an accelerator. It gets you past the maybe gate, where a project sits finished but unsigned

•    Testing non-deterministic AI means listening to every model on your network, mapping the regulations that apply, assessing compliance state, then sampling transactions for immutable logging. That is what Beacon does

•    The rush to AI dropped software fundamentals. Automated deployments, automated rollbacks, observability, and blast radius control are the highest leverage fixes for a stalled pilot portfolio

About Ken Johnston

Ken Johnston is VP of AI and Data at Envorso and a founder of the AI GovOps Foundation, a nonprofit practitioner community advancing AI Governance as Code. He spent 25 years at Microsoft and then led a Ford subsidiary serving 20 million connected vehicles, with senior roles spanning AI, data, cloud platform, telematics, and digital transformation. His book The Lean AI Handbook is due from Pearson in summer 2026. He is known for helping enterprises embed governance controls into AI systems through engineering rigor, automation, and operational feedback loops rather than policy documents alone.

About Bob Rapp

Bob Rapp is Principal AI Architect at the General Motors AI Center and co-founder of the AI GovOps Foundation. He has built AI and machine learning systems across automotive, healthcare, telecom, and industrial settings, with prior leadership roles at IBM Watson, Microsoft, GE Healthcare, and Vodafone, including some of the first models shipped to diagnose certain cancers. At the foundation he focuses on turning AI governance into an engineering discipline through governance as code, operational controls, and open source tooling.

 

In this episode

00:00 Welcome and guest introductions
01:45 Ken: GDPR, privacy, and the road to AI governance
03:32 Bob: when the models changed and the controls did not fire
04:50 What governance as code actually looks like
05:54 Where the policy code lives
07:06 Why gates also have to run at runtime
08:21 Will the CI/CD vendors build this?
09:20 Why the tooling is open source
11:07 Agent swarms, or viruses with credit cards
11:52 Ford: the puddle that flipped the car
14:10 GM: governance treated as a safety system
15:14 Over-the-air updates and automated targeting
16:07 Governance After Hours in San Francisco
16:42 The biggest misconception: governance as a brake
17:34 Unsafe at any speed
18:00 How much testing is enough
18:24 Red teaming and adversarial testing
19:14 The security analogy: shift left
19:53 Getting past the maybe gate
20:33 How many models do you test against
21:05 Inside Beacon
22:19 Umbrella, Lantern, and the audit layer
23:14 The Lean AI Handbook and the learning loop
24:47 Blast radius control and rollbacks
25:27 Data as the new oil, refined
26:54 Bob on where to start
28:23 Ken on his free e-book
29:26 Executive clarity and prototype theater
30:33 The one thing to remember
31:26 Wrap-up

In their words

“If your policies don’t compile, you’re not governing AI, you’re just hoping.”

— Bob Rapp (05:12)

“The most dangerous control is the one that’s stuck in Word or a PDF.”

— Bob Rapp (14:38)

“Ten thousand agents, or as I like to call them, viruses with credit cards, unless you’ve got the right controls.”

— Bob Rapp (11:07)

“Governance is an accelerator. It gets you past the maybe gate and into production.”

— Ken Johnston (20:30)

“Everybody in a rush to get to AI seems to have dropped many of the fundamentals of good software engineering.”

— Ken Johnston (23:40)

“Governance that lives in a PDF doesn’t run in production.”

— Bob Rapp (30:41)

 

Resources

Websites and projects

  • AI GovOps Foundation: (also aigovopsfoundation.org). Nonprofit practitioner community behind the open source projects, with both guests’ writing and social links

  • Beacon: Open source governance agent, live now. Signs the evidence

  • Umbrella: Compiles evidence into executable controls

  • Lantern: Reads the evidence back, so auditors and policy people who are not technical can follow the audit trail

  • AI GovOps Newsletter

  • Ken Johnston’s free e-book on AI governance fundamentals: Case study driven, free to download, on the foundation site. Third version in four months, so link the page rather than a file

  • Ken Johnston on LinkedIn

  • Bob Rapp on LinkedIn

Books

  • The Lean AI Handbook, Ken Johnston: Pearson, summer 2026. Applying software engineering fundamentals to AI: agile development, incremental deployment, blast radius control, observability, and the learning loop

  • Unsafe at Any Speed, Ralph Nader (1965): Referenced by Bob Rapp for enterprises that fill out forms and hold retreats but have no governance they can execute

Also referenced

  • Governance After Hours, San Francisco: Co-hosted by AI Realized and the AI GovOps Foundation. This episode builds on that room

  • GDPR: The standard that started Ken Johnston on the path from privacy to data governance to AI governance

  • Zoox: Autonomous taxi case study Ken Johnston cites alongside the Ford telemetry example

Related AI Realized episodes and events

 

Frequently Asked Questions

 
 
 
 
 
 
 
 
 
 
 
Previous
Previous

Your AI Agent Is Not the Risk. Its Authority Is.

Next
Next

Artifact-Scoped Agents: Stop Mimicking Job Titles