Organize the vision (2c) — what to design (and what is garbage)

For James — from a product-design lens.

You’re not wrong to be angry. If someone “designed” this step without reading what Organize actually is in Quorum, you get bullshit: random dashboards, fake analytics, generic kanban, or “project management” widgets that are not the methodology. That isn’t “bad designers” in the abstract—it’s missing requirements. Below is the tether so Figma (or any designer) cannot freestyle.

0. Inputs carried forward (V1 contract — non-negotiable)

Organize does not start from a blank spreadsheet. The workspace must surface:

Initial designs from 2b — the baselined concept variation (previews/links) and any retained alternates the user kept.
Concept framing — direction titles and short what this option is copy so themes stay tied to intentional design choices.
Core needs from step 1 — problem, users, constraints, risks, and the capability / feature list the team already captured.

Implement as OrganizeContextStrip (or equivalent) above the theme board: collapsible OK; absent is not OK. Step 3 (three-pillar analysis) continues to use the same bundle so user + AI collaborate on features with designs + concept + needs in reach.

1. Job to be done (exactly one)

After ideation (1), concept alignment (2a), and concept visuals (2b), the user has a kitchen-sink vision—features, ideas, maybe chaos. Organize (2c) exists so the AI team clusters that into themed groups with dependencies; the user reviews, edits, and locks that structure before three-pillar analysis (3).

PRD: “AI team clusters and themes the kitchen-sink features into coherent groups — identifies dependencies, groups related capabilities, surfaces natural release boundaries. User validates the organization before the filter runs. Concept artifacts from 2b are grouped alongside structured data by theme.”

If the UI doesn’t make that obvious, the design is wrong—not “creative.”

2. What belongs on screen (required)

Area	Content	Why
Left rail	Full pipeline; 2c Organize the vision = current (see `ux-design-specification.md` §4.2.1).	Orientation.
Team room	Full team present (John + roster per PRD); short John framing: “We’ve grouped your vision into themes—tighten names, move items, or ask us to re-cluster before we go to three-pillar analysis.” User can ask for regroup, explain mistakes, challenge grouping.	Organize is still a conversation, not a silent form.
Center (primary work surface)	Theme groups = primary object. Each theme is a card or column containing: theme title (editable), list of capability/feature items (from structured vision data), dependency hints (e.g. “depends on Theme A” or inline badges), optional release-boundary hint (MVP vs later) if data exists. User can drag item between themes and/or edit text per FR20.	This is FR19 (proposal) + FR20 (edit/confirm).
Concept linkage	Per theme (or per item): thumbnails / links to 2b concept frames that belong to that theme—grouped alongside structured data, not in a separate silo.	PRD explicit.
Primary CTA	Lock organization (or Confirm & continue to three-pillar analysis) — disabled until user explicitly confirms (or show warning on first click).	Checkpoint before step 3; activation milestone.
Secondary	Re-run clustering / Regroup with note (sends instruction to agents)—async OK (NFR-P3).	User agency when AI grouped wrong.
Job status	If clustering is heavy: Queued / Running / Failed strip or row (NFR-P3)—never silent spinner forever.

3. What does not belong (reject in review)

Fake “UX research” filter chips such as Engagement, Trust, Friction (or any other generic heuristic buckets) as the primary way to slice ideas—unless those exact words are verbatim theme titles returned by Quorum’s clustering for this workspace. Default UI: theme names are specific to the product (e.g. “Onboarding & activation”, “Billing & seats”, “Integrations”)—never stock categories copied from random UX articles.
Generic “Filter” or mystery next steps on the left rail as substitutes for the real pipeline (see UX spec §4.2.1).
Conflict resolution / Sync as fake methodology steps.
Unrelated charts (velocity, burndown, random KPIs)—unless directly tied to “release boundary” or cost later (that’s step 6–7, not 2c).
Empty “widgets” with lorem ipsum “metrics.”
Final design-system polish—2c is concept-era; fidelity is organized structure, not marketing landing page.
Solo random PM copy with no tie to themes + dependencies + concept links.

If you see any of the above, send it back: “Not in PRD for 2c.”

4. Kinsley’s design intent (spatial + concept)

Kinsley co-owns how concept visuals sit next to structured themes—the user should see which rough screens belong to which bucket before the filter tears scope apart.
Visual weight: themes and items first; concept thumbnails support the grouping, not decorate empty space.
Enterprise trio: same surface; optional presence on who is editing (later)—don’t block V1 on perfect multiplayer.

5. Copy hooks (tone)

Header / step title: “Organize the vision” — subtitle: “Review how your team grouped features into themes. Lock when you’re ready for three-pillar analysis.”
Save point (after lock): Celebrate the activation line from PRD: user earned structured vision + concept linkage—not “task complete.”

6. Acceptance checklist (for you or QA)

[ ] User can see all proposed themes and items in one coherent view (scroll OK).
[ ] User can edit theme names and move items between themes (FR20).
[ ] Concept artifacts from 2b are visibly tied to themes/items (link or thumb).
[ ] Dependencies / release-boundary hints visible where data exists (FR19).
[ ] Lock clearly gates progression to step 3.
[ ] No invented methodology UI that contradicts PRD pipeline rail.

7. Handoff line for Figma

Paste on the Organize frame:

Step 2c — Organize the vision (Quorum PRD). AI-proposed theme groups of features with dependencies; user edits/moves items; 2b concept visuals grouped by theme; CTA Lock organization before Three-pillar analysis. Not a generic dashboard. Spec: ux-screen-spec-step-2c-organize.md.

When implementation starts, trace to FR19, FR20, NFR-P3, and Epic 3 stories in epics.md.