← Market Analyst · c4

PICKUP.md

c4 Pickup Note

Contents
  1. Where work stands
  2. The decision queue
  3. Suggested first hour at pickup
  4. Quick orientation

Parked: 2026-08-20 — Significant work completed (display refinement, urgency/positioning extraction, company pages, segment cards), but full re-extraction never executed. Note: c8.agent-pivot is now the priority and will supersede the re-extraction work in this cycle. c8's extraction under new vocabularies (use-case-centric) will render this cycle's re-extract work largely redundant. Consider closing c4 in favor of c8 unless specific display polish is needed before c8 lands.

Where work stands

ThreadState
Display audit✅ Completed — All pages/components reviewed; identified and fixed issues
Extraction evaluation✅ Completed — Reviewed processed JSON; identified gaps (speaker attribution, insights quality)
Prompt updates✅ Completed — Updated extract-call.md with new guidelines (#1-13); added urgency + positioning extraction
Urgency & timeline extraction✅ Completed — Schema (BudgetSignal, UrgencyDriver, Urgency), validation, dashboard visualization
Positioning resonance extraction✅ Completed — Schema (ValueEcho, Positioning), validation, dashboard cards
Company pages✅ Completed — /companies index + detail pages with calls grouped by company
Segment card entity lists✅ Completed — Collapsible lists (5-item default + "Show all") for companies/attendees/calls per segment
Full re-extraction❌ Not started — data/processed/*.json not deleted; no batch re-extract run

Key artifacts in cycle:

  • CYCLE.md Key Changes section (detailed list of what was built)
  • No separate thread docs — all work captured in CYCLE.md

The decision queue

Unlocking decisions:

  1. Should c4 continue or close? — c8.agent-pivot's re-extraction under new vocabularies (use-case-centric, new fields) will make this cycle's work partially obsolete. If c4 continues, it will likely be superseded by c8's extraction. Recommendation: Close c4 in favor of prioritizing c8.

Canon/detail decisions (if c4 continues):

  • Re-extraction timing: batch API recommended (~$0.82 for 22 calls). Wait until urgency/positioning prompts are stable (they are now).
  • Display changes: All completed. No further display work needed before re-extract.

Suggested first hour at pickup

If closing c4:

  1. Archive CYCLE.md → mark Outcome as "Subsumed by c8.agent-pivot; display refinements retained; re-extraction deferred to c8" (5 min)
  2. Close the cycle formally (5 min)

If continuing c4 (not recommended):

  1. Re-read CYCLE.md Key Changes section to reorient on what was built (5 min)
  2. Decide: full re-extract now, or wait for c8? (10 min)
  3. If proceeding: npm run extract:batch -- --force to run batch re-extraction (async, ~2 hours wall clock) (5 min)
  4. Monitor batch status; after completion, npm run index to rebuild (5 min)
  5. Verify dashboard displays new fields (urgency, positioning) correctly with fresh data (15 min)

Quick orientation

  • Code readiness: Display work is done and merged. Extraction prompt is updated and validates. Schema changes (urgency, positioning) committed. No uncommitted work.
  • Key files: schemas/call.ts (Urgency, Positioning interfaces), pipeline/prompts/extract-call.md (guidelines 1-13), site/pages/calls/, site/pages/insights.vue, site/pages/segments.vue, site/pages/companies/
  • Uncommitted work: None
  • Cost if re-extracting: $0.82 total via batch API ($0.037/call)
  • Blocking: Depends on c8 pivot decision. c8 will do fresh re-extract; c4's re-extract work likely redundant.

Qpoint Brand Style Guide