Skip to main content
FeatureFactory

Pillar 01 · Measure

Know whether AI-assisted delivery is actually getting better.

Measure the quality, speed, authorship, and durability of every shipped change-then trace each number back to the work behind it.

Read-only GitHub access Visible scoring rationale Confidence-aware attribution

AI-assisted vs human

Quality and flow comparison

Sample data
025507510082 AI-authored88 AI-assisted79 HumanAvg PR quality by author

What Measure gives you

A decision system-not another activity dashboard.

PR quality with a rationale

Score each merged change against a visible rubric for correctness signals, tests, scope, and maintainability.

You get: A 0-100 score you can open, read, and challenge.

AI attribution with confidence

Separate known AI-authored, AI-assisted, and human work using explicit repository and identity signals.

You get: An authorship view that distinguishes evidence from estimates.

Delivery flow in context

Track lead time, coding time, pickup, review, and merge without reducing engineering work to one speed number.

You get: The stage creating delay-not just the total duration.

Rework after merge

Follow churn, reverts, fixes, and related regressions after a change lands instead of declaring victory at merge.

You get: A durability signal that exposes false velocity.

Team and repo comparisons

Roll metrics up by team, repository, work type, and authorship while keeping the underlying changes accessible.

You get: Comparable trends with enough context to avoid leaderboards.

Decision-ready scorecards

Read quality, flow, authorship, and outcomes together for a release, initiative, team, or period.

You get: One view for deciding where to invest or intervene.

Inside FeatureFactory

See the evidence in the product.

These views use representative product data. Open any image at full size to inspect the workflow and supporting context.

Measure overview with delivery velocity, quality, and trend charts.
Track delivery speed and quality as durable trends instead of isolated activity counts. Open full size ↗
Pull request detail with score explanation, evidence, and AI contribution attribution.
Every score stays connected to the code evidence and attribution behind it. Open full size ↗

Example · evaluate an agent rollout

Find out whether the speed is real.

A team introduces an AI coding agent on a checkout service. Measure compares like-for-like shipped work before anyone declares success.

Illustrative workflow · sample data, not customer results

Baseline

01

Define the comparison

Choose the repo, work type, period, and existing quality and flow measures.

Sample: checkout service · feature work · previous 30 days

Attribute

02

Separate the work honestly

Use explicit authorship signals and show unknown or estimated work instead of forcing certainty.

Sample: 18 AI-assisted · 7 human · 3 unknown PRs

Decide

03

Read speed with durability

Compare cycle time, review effort, quality rationale, and post-merge rework on the same cohort.

Sample decision: expand on scoped UI work; raise review on payment logic

Decisions, not vanity metrics

Know what to do next.

Each view is designed to resolve a real product or engineering question-and keep the evidence close enough to challenge.

Where does AI help?

Find work types where AI-assisted changes move faster without lowering review depth or durability.

Where is risk rising?

Trace low-scoring or high-rework changes back to the diff, rationale, repository, and workflow.

What should we change?

Adjust agent scope, review policy, specification quality, or team process based on the actual constraint.

Proof you can inspect

The evidence stays one click away.

Before you connect

Questions teams ask about Measure.

Each change is evaluated against a visible rubric covering correctness signals, testing, scope, maintainability, and review context. The score includes a written rationale so reviewers can inspect why it moved.

See it on your own work

Measure one repo before changing the whole workflow.

Join early access to establish your quality, flow, and AI-attribution baseline from your own delivery history.

Read-only to startNo replacement rolloutYour IP stays yours

We’ll use this only to coordinate early access.

Your baseline comes from your data-not a generic benchmark.

Continue the loopNext: turn signal into a build-ready plan