-
Notifications
You must be signed in to change notification settings - Fork 3
Closed Loop Product Selection Workflow
DDClaw edited this page Aug 4, 2026
·
1 revision
Version: 3.3 | Validated: 20 rounds, 82.7% avg adoption rate | Panel: 7 independent sub-agent reviewers
The Closed-Loop Product Selection Workflow is Sorftime Seller Agent's flagship end-to-end product research pipeline. It covers the full journey from product discovery through supply chain sourcing, financial analysis, risk assessment, and a 7-member independent AI seller review panel that votes on every product — then delivers a complete Go/No-Go decision package with post-launch monitoring.
/goal /sorftime-seller-agent Execute Closed-Loop Product Selection for yoga mats on Amazon US. Seller: beginner, $10000, private-label.
| Path 1: HPI Product-First | Path 2: Market-First |
|---|---|
| Find undervalued individual products | Analyze categories first, then products |
potential_product → full HPI ranking |
category_report → 11-dimension scoring |
| Best for: beginners, <$10K, opportunistic | Best for: brands, factories, strategic entry |
| 🎯 Sniper: "one overlooked gem" | 🗺️ Cartographer: "best land to claim" |

| Phase | Name | Key Action |
|---|---|---|
| P0 | Seller Input Gate | Budget, platform, stage, model, path selection |
| PA | Product Discovery | HPI ranking OR market panorama → ≥10 products |
| PB | Deep-Dive Verification | detail + trend + traffic + reviews (parallel batches) |
| PC | Supply Chain | 1688 sourcing, full landed cost, first-order qty |
| PD | Financial Analysis | Full P&L → true net margin (NO verdict here) |
| PE | Risk Assessment | 4-tier + brand + seasonal + IP + compliance |
| PD2 | 7-Member Seller Review Panel | 7 independent sub-agents vote GO/CAUTION/NO-GO |
| PF | Final Deliverable | Decision table, risk registry, first-order plan |
| PG | Post-Launch Monitoring | /loop command with 30/60/90-day checks |
The verdict is NOT an algorithm — it's 7 independent AI sub-agents, each role-playing a different seller perspective:
| Seat | Role | Weight | Veto? |
|---|---|---|---|
| 1 | Peer Match (same stage/budget/model) | ×2 | No |
| 2 | Peer Alt (tighter budget, opposite risk) | ×1 | No |
| 3 | Mentor (one stage above) | ×1.5 | No |
| 4 | Conservative (risk-averse) | ×1 | No |
| 5 | Opportunity (growth-oriented) | ×1 | No |
| 6 | Platform Specialist | ×1.5 | 🚫 Yes |
| 7 | Financial Auditor (P&L-only) | ×1 | 🚫 Yes |
Voting: GO ≥4/7 with Seat6 ≠ NO-GO. NO-GO ≥4/7 OR Seat7 veto. Panel may override algorithmic thresholds.
-
product_traffic_terms: Useexposure_positionfield, NOTorganic_searched_percentage -
product_trend: Parameter isproduct_trend_type, nottrend_type - PD2: Must spawn 7 sub-agents. Main agent FORBIDDEN from voting.
- All data: [VERIFIED] / [ESTIMATED:formula] / [ASSUMED:source] / [UNAVAILABLE:reason]
- 🆕 Closed-Loop Selection ← Full pipeline
- 📖 Case Study: Water Bottle ← Real data example
- Loop & Goal Templates ← 58 automation recipes