← Notes

02 / Human attention

More comments can mean more work

A review agent also creates work for its reader.

· 2 min read

What the teams report

Uber found that a single prompt produced false alarms and valid comments with little value. Its uReview system checks confidence and removes duplicate comments. It also suppresses categories that engineers rarely use. [1]

“Precision Is More Valuable than Volume”

A section title in Uber’s uReview report. [1]

HubSpot made its reviewer faster, but review quality remained a problem. It added a judge agent to check comments before publication. [2]

01FindPossible issues
02FilterCheck the evidence
03ReviewHuman attention
Our simplified illustration. Both teams check comments before publication; their checks differ.

Our observation

Measure the work after the output

Comment count shows how much an agent writes. It does not show how much useful work the team completes.

A useful evaluation can track accepted findings, review time, and missed defects. A filter can reduce noise and still remove a real issue.

A model judge can also make mistakes. The reports do not establish that an extra agent always improves a review.

A question for your buildDoes each additional comment save more work than it creates?

Sources

  1. Uber: uReviewQuality filters, duplicate removal, and feedback from engineers.
  2. HubSpot: Sidekick code reviewThe change to Aviator and the judge agent.