Case studies
Work
What the studio has shipped, told from its own records. Every study cites its sources; every claim is sized to its evidence.
Updated
TL;DR
Each case study here is written against the studio's own records (run receipts, pull requests, decision records), and its claims are explicitly sized: a study says 'scoped' until the evidence for a bigger claim lands, then says so with a dated addendum. Premise-stage entries appear as premises, without links, until they publish.
On the rail
How KDC gets built — and maintained
One person and an auditable gnome workforce founded, populated, and now maintain a live, source-cited open dataset. Every run lands in the cost ledger; every change arrives as a PR.
The studio builds its own client portal, with its own pipeline
The studio's first client engagement is itself: a private client portal, built by the same brief-to-PR pipeline that shipped kdc, under full client-tier discipline. Premise-stage — facts accrue from the founding onward.
How a study earns its place here
Case studies move through gates (consent, scrub, client review, operator merge), and a stalled gate simply doesn't publish. Client work appears only by recorded consent, scrubbed, under its codename. The studio's flagship study is about the studio itself, because the honest first client for this machinery was us.
An honest negative is data too.