https://dvirsegal.github.io/tags/ai/Recent content in Ai on Dvir Segal's blogHugoenFri, 02 Oct 2026 20:32:36 +0000https://dvirsegal.github.io/posts/ai-approved-the-pr-nobody-knows-why-the-code-works-a7f71e96dd63/Fri, 02 Oct 2026 20:32:36 +0000https://dvirsegal.github.io/posts/ai-approved-the-pr-nobody-knows-why-the-code-works-a7f71e96dd63/<p>Picture this, you open a PR and the AI report is already waiting. Three findings, all minor. You skim the diff, approve, and move on. A month later someone asks why that service retries three times instead of failing fast. The author doesn’t know, because an agent wrote it. You don’t know either, and you approved it.</p> <div style="text-align: center;"> <img src="https://dvirsegal.github.io/images/ai-approved-the-pr-nobody-knows-why-the-code-works-a7f71e96dd63/2.png" alt="LEGO-style scene, one month later: a developer asks why the service retries three times instead of failing fast, while the author shrugs and the reviewer holds an APPROVED stamp" style="display: block; margin: 0 auto;" /> </div> <p>I review PRs with an <a href="https://gist.github.com/dvirsegal/987e1d5dbd423640e86d9102e80edf1c" class="external-link" target="_blank" rel="noopener">AI code review workflow</a> (gist) that maps the blast radius of each change with <a href="https://colbymchenry.github.io/codegraph/" class="external-link" target="_blank" rel="noopener">CodeGraph</a>, a local tool that turns the codebase into a queryable graph of callers and callees, and reports findings by severity. It catches real bugs. It can’t tell me why the code is the way it is.</p>