Review Budget
A PR attention-cost dashboard for engineering directors that scores every pull request by the minutes of human reviewer time it burns, flags bloated AI-generated comments, and shows where team attention is being hemorrhaged.
engineering directors managing review-heavy teams
- Per-PR attention-cost score based on diff size, comment count, comment length, and reader-time proxies
- Slop detector that flags verbose, padded, or AI-generated comments before they hit reviewers
- Reviewer-load heatmap showing who is overloaded this sprint and which bottlenecks are forming
- Weekly ROA digest calling out which PRs/comments were worth reading and which were not
A Dev.to post naming 'Return on Attention' as the real bottleneck just went viral, with a CEO pointing out that AI-generated text is now weaponizing teammate attention; engineering leaders are actively shopping for ways to measure and police it.
Multiple credible outlets (Codacy, RDEL, AI Policy Desk, Medium) confirm AI-generated PRs are creating a real review bottleneck and 'attention hemorrhage' problem — concrete signal beyond the weak 3-reaction source post.AI Is Breaking Code Review: How Engineering Teams Fix the PR Bottleneck ↗RDEL #129: What keeps AI-generated pull requests from getting merged ↗
EngViz already markets 'DORA metrics + PR analytics' for GitHub, a OSS PR Analytics Dashboard exists, and LinearB/Swarmia/Sleuth/Jellyfish dominate the PR-review-metrics category — incumbents cover most of what Review Budget proposes.EngViz - Engineering Metrics Dashboard ↗PR Analytics Dashboard - GitHub ↗
Proven B2B SaaS willingness to pay in PR/engineering metrics (LinearB, Swarmia, Sleuth, Jellyfish all charge per-seat), but adding another dashboard tool on top of these is a harder sell for budget-tight eng leaders.AI Is Breaking Code Review: How Engineering Teams Fix the PR Bottleneck ↗
AI-assisted code generation will only expand, sustaining the review-attention problem — but GitHub/GitLab native features or existing metrics suites could absorb the niche without a standalone product.The Dark Side of AI in CI: Faster Pull Requests, Slower Teams ↗
GitHub/GitLab APIs + comment classification are tractable, but accurate 'minutes of reviewer time' modeling and broad multi-repo rollup are non-trivial to ship reliably.PR Analytics Dashboard - GitHub ↗