For managers
The pre-AI baseline: the number your team doesn't have but needs
You can't measure AI impact without knowing what your team looked like before. How to establish the baseline, and what it changes once you have it.
For managers
Churn ratio: the quality signal most engineering teams aren't reading
What a healthy churn ratio looks like, what elevated churn suggests, and why AI-assisted commits are showing 2× the rate of human-authored ones.
For managers
How to have a better conversation with your team about AI tools
Measuring AI output is politically loaded. A practical guide to using data as a starting point for conversation rather than a verdict.
For managers
The difference between measuring activity and measuring output
Commit counts, PR velocity, lines added — none of these are quality signals. The distinction Scryable is built around, and why it matters for how you manage.
For managers
Your developers are using AI. Your code review process isn't designed for that.
Standard review practices were built for human-authored code. What needs to change when a significant portion of commits aren't.
For developers
What your git history says about you, and who gets to read it
Understanding your own patterns before someone else defines them. Your data, your narrative — what it means to own both.
For developers
AI pair programming: when it helps and when the diff tells a different story
An evidence-led look at where AI coding tools genuinely accelerate good work, and where the data shows them introducing noise.
For developers
Commit hygiene in the AI era
How AI-assisted commits change what good commit discipline looks like — message quality, granularity, and what reviewers can reasonably expect.
For developers
Why your velocity is up and your net lines added is flat
More code written, more code deleted, similar net change. The gross vs. net lines dynamic, and when the pattern is worth paying attention to.
Research & data
The GitClear data on AI code quality: what the numbers say
A careful reading of the GitClear longitudinal research. Optimistic but rigorous — what the data actually shows and what it doesn't.
Research & data
Why 63% of developers use AI coding tools but only 29% trust the output
The trust deficit in AI-generated code is measurable and widening. What drives it, what the data shows, and what measurement does to close it.
Research & data
What the 4× rise in duplicate code means for long-term maintenance
AI assistance is producing four times as many duplicate code blocks as before. The downstream costs of that — on review time, test coverage, and debt.
Research & data
What a healthy AI-assisted codebase looks like
Proposed benchmarks for churn ratio, duplication, and net lines that suggest AI tools are working well. A reference point the industry is missing.
Research & data
The teams winning with AI: three patterns from the data
Teams treating AI as a strategic priority see 40–50% higher velocity. What they're doing differently — and how measurement sits at the centre of it.
Product & philosophy
Engineering intelligence vs. developer productivity: why the distinction matters
Measuring busyness is the wrong frame. The argument for measuring output quality instead, and what that shift requires of the tools you use.
Product & philosophy
Why engineering tools still take six months to procure
Engineering tooling is hard to buy. What self-serve, privacy-first products change about that, and why the procurement cycle is itself a signal.
Product & philosophy
AI adoption without measurement: the evidence against gut feel
The long-form case for measurement as the responsible path. Structured around the proof points, no moralising — just the data and what it implies.
Product & philosophy
Why we built for git history and not for integrations
The architectural decision to read git history directly — and what it means for privacy, simplicity, and the trust of the teams using Scryable.
Product & philosophy
What 'engineering intelligence' should mean
A category-definition piece. How the term gets used by competitors, and the specific, defensible definition Scryable is built around.
No posts in this category yet.