Skip to main content
FeatureFactory
Free tool · runs in your browser

AI vs Human PR Analyzer

Score a pull request's quality and effort, then see how it stacks up against a typical AI-authored, AI-assisted or human PR - with a plain-English read.

The pull request

Author type

Scores, baselines and weights are illustrative, editable estimates - a fast sanity check, not a benchmark.

Analysis

0-100
82Strong
Effort estimate6.8 pts

Moderate - a normal review pass.

Focused diff (180 lines)+6
7 files touched (contained)+4
2 test files (light coverage)+8
4 review comments (engaged review)+4

vs a typical AI-assisted PR

Diff size vs typical-5%
Files touched vs typicalon par
Test coverage vs typical-11%
Review comments vs typicalon par

This AI-assisted PR is about the size you would expect for a ai-assisted PR; test coverage is lighter than usual for this author type; AI-assisted work still needs a human owner accountable for the outcome.

Comparing AI and human pull requests

As AI moves from autocomplete to authoring whole pull requests, the interesting question is not "did a machine write it?" but "does this change hold up like a good change should?" The signals are the same for everyone: a reviewable diff size, tests that exercise the change, a contained blast radius, and a review that actually engaged with the code. This analyzer turns those signals into a quality score and an effort estimate, then benchmarks the PR against a typical one of the same author type.

In practice, AI-authored diffs tend to run broader and lighter on tests, AI-assisted work sits in between, and human PRs carry more implicit context. Those tendencies are illustrative defaults here - the real value comes from calibrating against your own repo. To go deeper, see our measure AI-generated code approach and our AI coding metrics.

Whoever - or whatever - wrote it, the bar for merge stays the same. For the reasoning behind these weights, read our guide on AI vs human code quality and code review best practices, or explore the full set of free engineering tools.

Frequently asked questions

It turns a handful of PR signals - author type, diff size, files touched, tests changed and review comments - into an illustrative quality score, a rough effort estimate, and a comparison against a typical PR of the same author type. The weights and baselines are opinionated, editable defaults, not measured benchmarks.

Measure what your team actually ships

FeatureFactory tracks PR quality, effort and AI-generated code across your whole repo - the feature factory that measures what it ships.

Get early access