How to deal with tech lead that uses AI for PR reviews
retained text
I'm in a scrum team where we had a great tech lead previously who moved to a different team and we got a new tech lead a couple months ago who uses AI for everything. This guy burned through 2K worth of tokens a few days into the month! Usually, the token budget resets every month. He has not built anything at all. Mostly small PRs and config changes.
He stopped reviewing PRs and created an Agent to review the team's PRs on his behalf. Keep in mind that we already have a PR bot! Anyway, his bot would leave so much text on every single PR that is mostly AI slop and noise. He literally expects us to respond to each and every comment left by his bot.
At one point, there were 10 different blocks of text left by his bot on one of my PRs. The PR was a small PR to add 2 columns to a schema. This literally copy and paste from one of the PR change requested. It is only a small snippet:
Ticket alignment: fine. PR body already flags that and points to the real work.
Schema has been verified? The claim has no provenance. Treating it as acknowledged rather than hidden, but flagging since nothing enforced the ordering.
Major - (Needs a fix):
In Spark SQL, cd AS create_date , not two columns — AS is optional in the grammar, so this parses cleanly and fails silently. It doesn't corrupt anything, it just means the "no drift across the existing column list" claim isn't actually true for create_date specifically. Worth fixing and re-running before treating that table's validation as complete. Initially looked like it could cause false-positive drift.
The review is awful and unnecessarily verbose. Makes shit up sometimes too. He completely stopped manually reviewing PRs and would completely rely on the bot. He would block team's PRs until all bot comments are addressed. I have raised issues during team meeting that we already have automated AI code review and his bot is redundant. He refused and said the bot provides a different context.
Now, I am forced to use AI to respond to his bot's wall of comments but he won't manually approve the PR and we would have to satisfy his bot wtf. This is driving me up a wall.
The problem is that I get change requests which don't make sense in a weird convoluted sentence structure and waste time on pushing back. I also get NO proper PR review or feedback.document fetched 2026-09-03 00:26 · canonical https://reddit.com/r/ExperiencedDevs/comments/1w5ifk8/how_to_deal_with_tech_lead_that_uses_ai_for_pr