Skip to content

Journal

Eight weeks, one outcome

How a studio sprint runs — diagnostic, one live workflow, handoff. Not a programme in disguise.

Large firms sell transformation. We sell one production outcome in a timebox you can hold in your head. Eight to twelve weeks. Senior-led. On your stack. When it is done, something is live that was not live before.

Week zero is a diagnostic

Two or three weeks, fixed scope. We look at the system under strain — the P99, the RAG that never left staging, the ERP that bills from rumour. You get a written path and an anti-path: what we will not build. If the honest answer is “this is a brochure,” we say so and leave.

The sprint is one workflow

Not a platform. Not a five-year roadmap. One workflow: an on-call summarizer with evals, a gRPC path that holds an SLA, a trip event log that makes invoices true. The model, the service, the dashboard, the alert — whatever “done” meant in the diagnostic.

You get the principal, not a bench. Reviews, architecture, and the unglamorous jobs (idempotency, traces, runbooks) happen in the same seat.

Handoff is part of the work

We do not leave you a slide titled Next Steps. We leave you the workflow, the eval set, and the person on your team who can change it. Embedded principal is available after if you want the same hands on the next system. It is never “we have fifty developers ready.”

If you need a website, this is the wrong studio. If you need a system that will still be true in six months, eight weeks is long enough to tell.