What you walk away with
Every engagement ends in something tangible you own and can keep improving — not a slide deck. Usually one or more of the three below, and always built so it stays changeable long after we’ve gone.
Output one
A working service
A validated MVP that flexes into production — built lean in Django and HTMX on standard Postgres, or n8n, or both. It runs as a web app and installs as a PWA, now a sensible default on desktop as well as mobile; and where a service genuinely needs to be in the app stores, we wrap the same codebase as native iOS and Android apps with Capacitor, rather than maintaining a second, separate build.
Services can carry what modern public services actually need — mapping and geolocation, offline use in the field, and AI woven into the workflow rather than bolted on the side. All of it on deliberately boring, open tools a future maintainer already knows, human or AI: no exotic dependencies, no lock-in, nothing you can’t leave with.
Output two
The operating model
Service blueprints and process maps that capture how the service actually works — the steps, rules, decisions and exceptions, and the roles and responsibilities that go with them — in structured, human- and machine-readable YAML. One source of truth a person can read in a review and an AI agent can execute without ambiguity, instead of the real process living in one person’s head. Sometimes this alone is the engagement: the design, ready to run, without a line of bespoke code.
Output three
Living documentation
User stories, runbooks, blueprints and process maps kept current with every change — updated in the same change as the thing they describe, never written once at the end by someone tired. It’s what lets you verify what the AI built without reading the code, and it’s the memory the service carries forward: the reason a decision was made, not just the fact of it, so the next person — or the next AI agent — inherits the thinking, not just the artefact.
It comes with an honest roadmap, too: the product, technical and service work it would take to grow the service further, so you can plan and fund the next step with your eyes open.
Why they last
Put them together — a conventional, open stack, one readable source of truth, and documentation that stays true — and you get a service a person or an AI can pick up years from now and change with confidence. No black box, no bespoke framework only we understand, no lock-in.
That’s the quiet promise under the fast delivery: not just a working thing now, but a thing that keeps working — and keeps improving — long after we’ve gone.