Extracted abstract or opening context
**Evo-Cubed | Research Engine Output** **Date:** 2026-03-19 **Topic:** What does it take to systematically capture a US city for a plant-based oat milk brand, city by city? **Prior Evolutions Referenced:** 4 (koatji-gtm, koatji-miami-beachhead, koatji-outreach-v3, koatji-gtm/path-c-iaas)
The Milkmen/Koatji operation has been built across 4 prior Evo3 runs. The current state as of 2026-03-19:
| Asset | State | Evidence | |-------|-------|----------| | Prospect database | 3,500+ records across 22 cities | Supabase `wovknbcvxjvayfdflxkn`, `sweep_prospects` table | | Outreach Command Center | 16 files, ~4,978 lines, 10-tab React shell | `OutreachCommandCenter.tsx`, `useOutreach.ts`, `GuidedSendPanel.tsx` etc. | | Mail Bridge | Python HTTP server, AppleScript Mail.app, GPT-5-mini classifier | `mail_bridge.py` (1,348 lines), LaunchAgent `com.koatji.mail-bridge` | | Resend Email Pipeline | 6 Edge Functions (send, batch, follow-up, tracking, unsubscribe, AI-generate) | Supabase Edge Functions on milkmen project | | Pipeline Automation | 9-phase sweep/enrich/score/email flow | `koatji_ig_pipeline.py`, 3 Supabase edge functions | | NYC Field Routes | 25 curated shops across 5 walking routes (Midtown) | `koatji-field-routes-midtown.txt` | | Warm Lists | SFL, NYC, LA CSVs (headers only -- data in Supabase) | `koatji-sofla-field-list.csv`, etc. | | Graduated Send Cadence | 10/25/50/100 per week scaling | `outreach-config.ts` | | Sequence Engine | 3-touch default, JSONB steps | `email_sequences` table, `prospect_sequence_state` | | Accounts Pipeline | potential -> high_interest -> sampled -> committed -> odeko_setup -> first_order -> reorder | `AccountsTracker.tsx` |
| Metric | Count | Source | |--------|-------|--------| | Emails sent (total) | ~97 | outreach-v3 Stage 0 (campaign start 2026-03-16) | | Responses received | 15+ | outreach-v3 Stage 0 | | Response rate (where measured) | ~16% | outreach-v3 Stage 0 | | Committed accounts (hot/interested) | 7 Miami accounts | outreach-v3 Stage 0 | | Named committed accounts | Caracas Bakery (3 doors), Mane Coffee, VI Coffee Bar (sampled) | Prior evolutions | | Current daily send limit | 10/day (Week 1 cadence) | outreach-config.ts | | Pipeline stage distribution | Unknown exact, mostly potential/high_interest | AccountsTracker pipeline |
| City | Prospects | HOT | WARM | Pipeline Status | |------|-----------|-----|------|-----------------| | Los Angeles | 1,833 | 0 | 0 | IG scoring in progress | | New York | 1,015 | 15 | 154 | IG complete, no Prospeo | | South Florida (aggregate) | ~500 | 134 | 175 | COMPLETE | | Fort Lauderdale | 52 | 18 | 24 | Complete | | Miami | 44 | 18 | 10 | Complete | | Doral | 45 | 11 | 15 | Complete | | Brooklyn | 37 | 0 | 0 | No ICP scoring | | Other cities (18) | ~400+ | minimal | minima
Why this is not always a full paper yet
Corpus pages are public-safe readers for discovered workspace artifacts. They are not automatically final papers. A corpus item becomes a polished paper only after the editable source, evidence checkpoints, references, figures, render path, and release status are attached through the paper schema.