Turn research into an implementation roadmap

Convert this notebook into an implementation roadmap for a small team that has to actually ship the thing the sources describe.

Produce:
1. End-state in one sentence — what "done" looks like, grounded in the sources.
2. Milestones — 4–6 named milestones in execution order. For each: the outcome, the sources that justify it, and the single thing that proves the milestone is real.
3. Dependencies — for each milestone, list what must be true before it can start. Highlight the critical path explicitly.
4. Workstreams and owner archetypes — 3–5 parallel workstreams with the kind of person who should run each (e.g. "infra-leaning engineer", "design-partner manager"). Don't invent names.
5. Risks and kill criteria — the 3 risks most likely to derail the roadmap, and the observable signal that should trigger a stop or pivot for each.
6. The smallest possible week-1 commit — one concrete action that could begin tomorrow and would still produce useful information even if the whole roadmap is later scrapped.