Track 10 · Discovery & handoverLesson 12 of 12
From one site to many
Turn site differences into configuration, onboard each site on evidence, and know what to productise.
- The FDE path · Stage 6 of 6
- 35 min read
- 75 min hands-on
- Pro
What you'll be able to do
01- Find where a new site differs from the first one (process, data, languages, regulation, volume and people) from evidence, not from the site's description of itself.
- Draw the line between configuration and code, and prove a new site needed no code.
- Run a site onboarding playbook: data contract, mapping, per-site golden set, shadow mode, success criteria, cutover in waves with stop rules, and hypercare.
- Judge each site's readiness on its own evidence: its golden set, its variants and its unit cost at its own volume.
- Decide what to harvest into the product, what stays manual, and when to stop customising.
Before you start
- The operational model: objects, links and actions: object types, properties, mappings and actions as the only write path. A site's configuration maps its columns onto that model.
- Shadowing the work as it's really done: event logs, variants and rework loops. Here, comparing them across sites.
- Golden datasets and the metrics that matter: what a golden set is and how to build one. Here, one per site.
- Canary releases and staged rollout: waves, rollback triggers and kill switches. Here, applied to sites.
- Python at the level of reading a few hundred lines. The worked example runs offline.
Starting here? On the core path this lesson comes after the operating rhythm (Rhythm, writing and people on site) and before handover (Handover, run state and the next use case). Two words used below: a site is one desk with its lanes, people and systems; the reference site is the first one live, which every other site is compared with.
The short version
The first site proves the agent works. The second shows whether you built something that can run anywhere else. Sites differ in their process, their data, their languages, the rules they work under, their volume and their people. Most of those differences should land in configuration: data with a schema that the product already knows how to read. A few need code, and some should stay with people. Onboard each site with the same playbook: a data contract, a mapping onto the shared model, a review of the site's process variants, a golden set built from the site's own cases, shadow mode, cutover in waves with stop rules, then hypercare. Judge each gate on that site's evidence. When two sites need the same thing by hand, build it once into the product. When only one does, keep it manual and say why.
- One product, many configurations. A site is the reference site's configuration plus its overrides.
- Configuration is data with a schema. If it needs a new kind of rule, a new action or a new object type, it is code.
- Prove it. A new site that needed no code leaves the product's fingerprint unchanged.
- Every site gets its own golden set, and the reference site's set runs on every change.
- Harvest at two. A pattern at one site stays manual; at two it becomes a product candidate.
Pro lesson
The rest of this lesson is in Pro.
The mental model, worked example, field note, failure table, exercise and self-check, with the voice tutor.
Pro opens this stage
Stage 6 · Ship, run and hand over- 8 lessons, with the voice tutor
- Lab · Check a trace export before you trust the dashboard, graded here
- Halden week 7 · The champion moves
- Halden week 8 · The portal goes down before cut-off
- Halden week 9 · Prices up, and a new rule
- Halden week 10 · A second site, and handover
- The lab kit, to run Halden weeks on your machine
Also in Pro
- Every stage and specialisation, in full
- Every lab and every role project
- Halden weeks 2–10, on your record
- Every drill, live in the playground
- All solutions, plus the server judge
- Lab Assistant and every Teach with AI