It’s wrong when devs use AI in secret.

In most cases, that even goes against the contracts you already have.

Instead:

• If management is hesitating, I’d start by pitching the idea to them.

• If devs are hesitating, I’d pitch it to the devs. 😀

The point is:

Agentic coding (not “vibing” with AI) is one of the biggest discussions on social media right now, and for good reason.

It makes sense for leaders to sit down with their teams and create a clear plan.

It makes sense for developers to understand the strengths and limitations of tools like Claude Code.

It does not suddenly disappear; it's in fact an ongoing evolution; a bit bumpy, yes, but an evolution nonetheless.

What I’m urging is simple: talk openly about AI usage instead of letting it happen like this:

• Everybody denies using it.

• Everybody starts using it in private side projects.

• Everybody continues denying it, but secretly uses it in the company codebase.

• Output and quality suddenly change in unpredictable ways.

• Developers suddenly get creative about what not to commit to git. :)

I’ve now seen this exact pattern play out in three different companies, and it must not be like that.

—Adrian