Accountability

As an AI-enabled software engineer, now more than ever (although this has always been true), the value you provide to your team is not the number of lines of code you generate write.

Also, no matter how great an agent is at producing code, the CEO of a company isn't going to spend all their time prompting the agent to code for them. They have other things to do (at least, the good ones...).

Click to play video

You do offer value in your:

• Ability to understand the code the agent produces – not just what it does, but why that's a good way to do it.
• Expertise to protect the codebase from vulnerabilities.
• Judgment to know when to push back on a feature because it will add too much complexity/latency/cost for the value it provides.
• Time and care that you put into prompting, re-prompting, and fixing AI-generated code, and ensuring its safe rollout.

Perhaps most importantly, you provide your own end-to-end accountability for the code you ship.

It's Not How the Sausage Is Made

It really doesn't matter to your users or your team how each individual line of code in your PRs was written.

• Did you type it at 150 wpm in Neovim?
• Did you copy/paste it from Stack Overflow?
• Did your agent generate it for you?

No one cares.

The bottom line is that the code is your responsibility, and saying "The agent wrote it" is not an excuse for bad code. In fact, it's a worse excuse than "I wrote it" because if you wrote it, at least you were trying. I'd rather have a team member who tries and fails than one who wants to be lazy and redirect blame.