ib.digital

Case studies

Proof, not pitches. Named in dollars.

Every case below carries a named outcome — dollars saved, hours removed, headcount replaced, or weeks shaved off a shipping target. Anything we can’t name in those terms doesn’t make this page.

2 published

Product · Built with AI

Prime Resume — an AI resume optimizer shipped in weeks.

Our founder's own AI product, built almost entirely with Claude Code and modern AI tooling. From blank repo to live beta in dramatically less time than a traditional build.

Shipped in weeks, not quartersRead case study

Automation · Medical EMR

UAT automation that replaces two QA engineers.

Claude Code + Playwright agents running the full test suite on every release — catching regressions a manual pass would miss.

~2 QA FTEs of cost removedRead case study
playwright · claude-code
spec: auth · login412ms
spec: auth · logout238ms
spec: billing · subscribe764ms
spec: billing · upgrade519ms
spec: onboarding · steps681ms
5 passed · 0 failed · saved ~2 QA engineers of manual effort

What we publish

Three rules, strictly applied.

A named outcome, or it doesn't ship.

Every case study includes the dollar, the hour, or the headcount. If we can't name it, we don't publish it. That rules out a lot of plausible-sounding work — and that's the point.

Production environments only.

Pilot demos and prototype videos don't count. The agent has to be running on a real customer's stack, against real traffic, before we put it on the page.

Founder on every project.

Every engagement is directly supervised by the founder alongside the senior team — discovery, build, and case-study write-up all run through the same principal. No account-manager layer, no handoffs to people you never met on the first call.

Let’s talk

30 minutes. Bring one problem.

No slide deck, no discovery call fee, no ‘request a quote’ games. We’ll tell you whether it’s a fit in the first call — and point you somewhere useful if it isn’t.