I'm Andrew, founder of Koalr — AMA about what actually predicts production failures
Hey PH 👋
I'm Andrew, founder of Koalr — a deployment intelligence platform that scores every PR 0–100 for production risk before it merges.
The idea came from 20 years in IT consulting and delivery. I've sat in more post-incident reviews than I can count, and the pattern is almost always the same: someone merged something on a Friday, nobody flagged it, and the team spent the weekend firefighting. The data to catch it was there — it just wasn't connected.
So I built Koalr.
What I can go deep on:
The 36 signals we use to predict deployment failures — what made the cut, what didn't, and why
How the model auto-tunes from your actual incident history
Blast radius propagation — modeling risk across service dependency graphs
CODEOWNERS governance and why drift is a hidden risk signal
Coverage-to-risk correlation (it's not what most people think)
Incident autopsy and what patterns we see across orgs
Why DORA metrics measure what already happened, not what's about to
Building a risk model without access to source code (we never read your code — only metadata)
The hardest signals to get right (spoiler: author expertise is deceptively tricky)
What it's like building a solo founder SaaS in this market
Try it before you ask:
No-signup demo: https://app.koalr.com/demo
Risk calculator (runs in browser): https://koalr.com/tools/deploy-risk-calculator
Ask me anything — technical, business, product, or "why does this signal matter." I'll be here all day.
Replies