Managing An AI Teammate: Lessons Bill Davy’s Journey to 99% Automation

🎧 Listen to the Audio Version

‍

Property managers tell us the same thing again and again: they’re tired of hearing about AI’s possibilities, they want to see the process.

The idea for this entire breakdown started with a question from a voice in the industry we really respect: Jen Ruelens.

While Jen isn't a client (though we'd love for her to be someday!), her Hold It with PM Jen- YouTube channel content is top-notch. She has a knack for asking the exact questions other operators are thinking and she recently put the core challenge of AI adoption best when she asked:J

“What does a company onboarding an AI agent successfully look like? Who on the team is supervising? How do they talk about it, how do they train their team on it?”

We believe many others feel just like Jen: excited about AI, but hesitant because the day-to-day still feels like an unknown.

That’s why we sat down with Bill “The AI Czar” Davy. When a client reports a week with 99% maintenance automation (meaning 99% of his work orders were coordinated end-to-end by AI without human involvement), our whole team stops and takes notes. That client is Bill Davy.

Bill isn’t a technologist; he’s an operator who walked into a new industry, faced heavy skepticism, and still managed to hit weeks of 99% maintenance automation.

We knew we had to understand his playbook. So, our leadership team invited Bill to a private Q&A at our recent strategic offsite, and what he shared was too valuable not to pass along. Here’s what we learned from Bill about the real process of onboarding AI as a teammate and the "nuts and bolts" that answer the biggest questions we hear: Who supervises the AI? How do you train the team? And how do you check its work?

It starts with Lesson 1: finding the right person to lead the charge.

Mini Ad:

Want to hear how Lacy Hendricks is paying $8.30 per signed PMA off SEO? Don't miss this.

Finding Someone (Like Bill) To Manage The AI

The first question every leader asks is: “Who’s actually going to manage this?”

I thought the answer was obvious: someone with deep property management experience. But Bill helped me see that the right champion isn’t defined by a resume,  it’s defined by mindset and skillset.

1 A Mandate to Disrupt. My first assumption was that you'd need deep property management experience to lead an AI initiative. Bill's journey suggested the opposite. The key quality wasn't his existing knowledge; it was his mandate to be disruptive.

He wasn't hired for his property management knowledge. He was hired for his willingness to challenge the existing process. He pointed to a pivotal conversation with his boss that illustrated this:

"I just asked him... 'Did you hire me because I'm willing to be disruptive?' And he flat out told me he hired me because I was willing to be disruptive."

That mandate to disrupt, he explained, was the foundation of his success.

2 An Operator's Skillset. The second pattern that clicked for me was that Bill’s background wasn't in tech; it was in operations.

He connected his success directly to his time at Lowe's, where he mastered two key skills:

• Hands-on Ops Management: He had an instinct for streamlining AI’s day-to-day tasks.  
• Process Optimization Under Pressure: He was used to spotting the small inefficiencies that create huge bottlenecks (a skill he immediately used to improve the AI's performance).

This is why he succeeded. When others see AI as a gadget to be installed, Bill approached the AI like an operations hire to be trained.

That simple shift in perspective is what separates a frustrating tech project from a 99% automation rate.

But here's the unexpected lesson he taught me: once you find that champion, their first test isn't with the AI. It's with their own team.

For Bill…Someone Has to Own It (The Human Hurdle)

AI onboarding rarely fails because of the tech, it fails because no one takes ownership. Success starts when one person steps forward and carries the responsibility while everyone else gets comfortable.

That’s what Bill did.

When most of his team resisted Vendoroo, he didn’t argue. He absorbed the risk himself and kept the AI’s training in his own hands. His teammates weren’t asked to train it; their only role was to bring issues to him so he could manage the AI directly.

‍

The insight here is simple: when most people convince everyone in the team upfront, Bill thought of one thing: it requires one person willing to own the outcome and give the AI a clean runway to prove itself.

With that foundation in place, the focus shifts from team resistance to the real work: hands-on training. That’s where Bill’s hands-on playbook makes the unknown visible.

Train Your AI Teammate (Bill’s Hands-On Playbook)

For months, I've heard people talk about "implementing" AI. The word always felt a bit cold, a bit... technical. It implied a clean process, like installing software. But I knew the reality had to be messier.

The conversation with Bill is what finally gave me a better framework. He doesn't install AI. He onboards it.

Immersion Beats Delegation

The first thing that clicked for me was his approach to the "black box" problem. How do you learn to trust something you can't see? Bill's answer was simple: you don't. You make it visible.

For two weeks, he became what he called an "overprotective dad," living with the AI side-by-side and reviewing every single work order.

He wasn't just watching; he was actively learning its logic and finding ways to optimize it on the fly, like when he tweaked the vendor assignment flow and immediately cut his cycle times in half.

That’s how you turn a scary unknown into a knowable process.

This intensive, side-by-side supervision is how you start. You don't hope it works- make sure of it. It's not just about catching mistakes; it's about actively teaching the AI the invaluable experience that lives in your head.

Encoding Your “Gray Area”

Another shift in my thinking came when I realized he wasn't just teaching the AI a process. He was teaching it judgment. We all have this "gray area" knowledge (the unwritten rules) that makes us good at our jobs. I never thought an AI could learn that.

Bill showed me it could.

He gave a perfect example: when his regular HVAC techs were booked during a heatwave, he used a trusted handyman. But then he did something crucial: he went back and taught that creative solution to the AI, adding the handyman to the emergency list. He was literally encoding his experience into the system.

That single action captures the essence of training: you’re not just fixing today’s problem, you’re building tomorrow’s playbook. And if you stick with it long enough, something powerful happens…the AI begins to think like you.

The Moment of Earned Trust

This led to the final, most important realization. Trust isn't a leap of faith you give out. It's a reward the AI earns.

For Bill, that "Aha!" moment came when the AI correctly diagnosed a complex leak as an HVAC issue…a call he admitted he might have missed himself.

That was the moment the "intern" graduated to a trusted teammate.. The scary idea of "an AI agent doing things independently" became a powerful, helpful reality. It’s the proof that the training is working.

This hands-on playbook is what takes you from fear of the unknown to real trust. But what does it look like when the training is done? What does a company that has successfully onboarded an AI agent really look like?

The Payoff: What “Done” Really Looks Like

Bill’s story gives us a clear and powerful definition of "Done," and it isn't about the AI being perfect. "Done" is when the AI is trusted to run the standard, day-to-day operations autonomously, freeing you to become a strategist. Your role fundamentally changes from a firefighter putting out daily blazes to an architect preventing them from starting in the first place.

This new reality isn’t a "set it and forget it" task; it's a “teach and scale” process.

The intense, daily supervision evolves into a collaborative rhythm of feedback and improvement. The AI teammate works 24/7 to triage and troubleshoot within the rules you set, while you provide the strategic oversight to make it smarter. As Bill discovered, this consistent feedback loop is what turns one-off fixes into a repeatable system.

Ultimately, the proof of a successful onboarding is in the results. The metric tells the story. When Bill hit a week with 99% maintenance automation, it was undeniable proof that the process works.

That is what "done" looks like: the unknown becomes your greatest operational asset.

Bill’s story shows what’s possible when you commit to the process. But his 99% automation week isn’t just his win; it’s a preview of what’s available to any operator willing to take the same steps.

Now It's Your Turn…

Bill Davy proved what's possible. We're looking for the next "AI Czar."

If you're that person on your team, reply to this email. We're assembling a group of the industry's next leaders.

If you're ready to put Bill's playbook into action, come see what the only AI-native maintenance solution can do.

If you want your own ‘99% week,’ it starts the same way Bill started: with the first step. Schedule a demo with Vendoroo.

Let’s build the future together.

—

Pablo Gonzalez,

Chief Evangelist at Vendoroo

‍

‍