Why AI Is Redefining Code Reviews — Ankit Jain Explains

Ankit Jain reveals how AI now writes and reviews code, but teams are losing the human context that made reviews valuable.

Ever wondered why code reviews are slowing down even as AI writes more of our code? The answer is both surprising and urgent.

⚡ TL;DR: AI is taking over both writing and reviewing code, but teams are discarding the valuable context that made reviews useful.

The Simple Version

Think of code review like a group chat that catches mistakes before they ship. Ankit shows that the chat is being replaced by silent AI checks.

How It Actually Works

Anatomy of a New Review Flow

🎯 Think of it this way: The AI writes code, then an AI reviewer comments, and a human skim‑feeds the thread before hitting merge.

From Session to Acceptance Criteria

The conversation is captured, turned into clear acceptance rules, and fed into a test plan that runs against a live preview.

Technical Deep Dive

Ankit introduces an AI slop registry that stores recurring review comments as reusable acceptance criteria. He pairs this with a deterministic test generator that validates intent rather than diff.

Why This Matters

When reviews lose their human nuance, teams miss mentorship, architectural insight, and onboarding moments. Preserving intent in prompts turns reviews into evidence that can be automated.

This shift rewrites the trust model: correctness is no longer the only goal; knowledge transfer must survive the automation layer.

Key Takeaways

AI writes code, AI reviews code, humans skim threads.
Half of review value is knowledge sharing, not just bug detection.
Specs written up front without feedback loop are just waterfall.
Prompt decisions become the real acceptance criteria.
Capture sessions to build an AI slop registry.
Generate test plans from captured intent for live previews.
Review surface moves from diff to intent and evidence.