Back in January I said I was going to build an AI working charter with my team. Here it is.
This wouldn’t be a policy or some dictated list of rules from on high, but rather a living document we would write together about how we were going to work alongside AI coding agents as a team, what we valued, and what guardrails we were putting around ourselves and the agents before we discovered after the fact that we needed them.
We did it. And there were a few things that surprised me.
The first surprise was where the pushback came from. I figured I’d be the one pushing for the “don’t let your skills atrophy” section, but I was wrong. In our first open discussion after presenting my engineering team with the draft, multiple of them brought it up. They wanted explicit, fleshed-out language about keeping their technical skills sharp and about not outsourcing deep understanding. This is because they got it intuitively: if you can’t read the code and understand what it’s doing, you can’t validate the output. And if you can’t validate the output, you’re not an engineer anymore. You’re just a person who ships things they don’t understand, and that is a recipe for disaster, not only at the enterprise level, but at the individual side project level as well. We’ve all seen plenty of “vibe coded” app disasters sprinkled through our LinkedIn feed at this point.
That section ended up being one of the most developed in the whole document.
The second thing that stood out was the healthy skepticism my team wanted baked in around the tools themselves. Not cynicism, but skepticism. There’s an important distinction there. The consensus in the room was that these tools will absolutely succeed or fail based on the architecture, guardrails, and conventions you build around them. The models and the tools are not the secret sauce. How you harness it is. So we put that in writing too.
What I actually came away with was something I didn’t expect going in: my team is thoughtful. Not just technically, but philosophically. They’re not naively excited but they’re also not dismissive. They’ve thought about what it means to be a software engineer right now, and they had opinions worth hearing.
The charter is a living document. We’ll revisit it and revise it regularly. But it exists, and we built it together, and I think that it’s the conversation around it that matters even more than what’s in it.
I’m sharing that AI Working Charter with you, because when we were writing this, I could not even find a decent example anywhere to use as a starting point. Hopefully our attempt at a charter can serve as that jumping off point for you and your team: read it here.
More to come on where we’re taking this next.