The end of LGTM.

Code review used to be where a senior engineer slowed down and looked. Slow on purpose. The one step that asked: does a person, not a process, think this should ship?

For thirty years that question got answered honestly, because writing code was the bottleneck. Reviewing a diff took less time than producing it.

Not anymore. Code is the cheap part. A thousand lines of plausible TypeScript takes less time than a coffee. The diff outpaces the reader.

Teams that adopted AI assistants didn't start reading more carefully. They started approving faster. The bar drops to meet the volume.

We refuse to accept that.

Review has to be done by a human, eyes on the file, for time that scales with the size of the change. That should be measurable. And the measurement belongs on the PR, next to the green checkmark, not in a compliance dashboard.

That's the whole product. A required CI check that turns green when a person has read the diff. No AI summaries. No auto-approval. No suggestions.

It's for the senior engineer who is the last one still opening the files. The tech lead who wants approval to mean something. The security lead who needs an audit trail that isn't a screenshot.

What we won't do.

• Read the code for you. That's the human's job.
• Track keystrokes, mouse paths, or screen contents. Dwell time on a file, nothing else.
• Frame this as catching anyone. Reviewers prove they read it.
• Hide pricing behind a sales call. It's on the page. Public repos are free.

What we will do.

• Open source the extension so anyone can audit what it sends.
• Publish a real changelog. Ship what's on it.
• Track reading, not readers.

Real review, on the record.
The Read PR team