Momo eCommerce & Live Innovation PoC
Published 2026-05-14Updated 2026-07-024 min read
"An eCommerce innovation PoC that ties massive SKU, member, live-broadcast, campaign, and in-house logistics data together with external trend, weather, economic, and competitor signals — solved through Ontology + Agentic AI."
Momo (富邦媒 / momo.com)'s millions of SKUs, TV home-shopping and live commerce, mobile app, and in-house logistics (24-hour delivery) consolidated into a single KG. Public information only.
🚀 Interactive Demo
👉 Open the demo page (new window)
Hands-on with 11 scenarios plus the core UIs for live-commerce attribution, delivery SLA, and recommendation diversity.
1. PoC One-Line Summary
| Axis | Content |
|---|---|
| For whom | 5 departments: Marketing · Category · Search/Recommendation · CS · Logistics |
| What | Massive SKU + Member + Live broadcast + In-house logistics + 4 external sources |
| How | Ontology (KG 25 classes) + Agentic AI |
| Differentiator | Live commerce attribution + Delivery SLA monitoring + Recommendation diversity |
2. Business Structure (Public Information)
| Axis | Content |
|---|---|
| Channels | Mobile app (#1 traffic in TW) · Web · TV home-shopping · Live broadcasts · Line/KakaoTalk |
| SKUs | Millions (long-tail strength) |
| Logistics | momo Fast (24-hour delivery), in-house fulfillment centers |
| Parent | Fubon Financial subsidiary |
3. Five-Department Personas
| Code | Department | KPI |
|---|---|---|
| P1 | Marketing (App · Campaigns) | App ROAS · Live concurrent viewers and conversion |
| P2 | Category · Trend | New-product trend SOV · Category LTV |
| P3 | Search · Recommendation · MarTech | Recommendation accuracy · Diversity · Search latency |
| P4 | Customer Service · CRM | NPS · Refund rate · Delivery satisfaction |
| P5 | Logistics · MD | Delivery SLA · Per-region fulfillment load |
4. 11 Scenarios Summary
| # | Scenario | Data Mix |
|---|---|---|
| S1 | Natural-language semantic search | SKU + Member + Live review + External |
| S2 | 5-department persona chatbot | All tools autonomously invoked |
| S3 | Category · BU insight cards | In-house GMV + Trend + Competitor |
| S4 | Persona matching + Clustering | RFM + Live viewing + Social |
| S5 | Omnichannel campaign ROAS simulator | Campaign · Live · App · Messenger |
| S6 | External signal fusion | 4 sources (Social · Weather · Economy · Competitor) |
| S7 | Omnichannel member journey | App → Live → TV → Re-purchase |
| S8 | Marketing consent · PII guardrails | Bedrock Guardrails + Ad-copy review |
| S9-Mo ⭐ | Live-commerce broadcast attribution | Viewer → Cart → Purchase funnel |
| S10-Mo ⭐ | In-house logistics delivery SLA + Order optimization | 24-hour promise vs measured |
| S11-Mo ⭐ | Mobile-app recommendation · Search diversity | Diversity · New-product exposure · Long-tail |
5. AWS Stack (Common)
Same as LG H&H + live-stream attribution (Kinesis), delivery SLA (TimeStream), ~800K edges (massive SKU + live events).
6. Top-5 Demo (30 minutes)
| Order | Scenario | Time |
|---|---|---|
| 1 | S1 Semantic search | 4 min |
| 2 | S9-Mo ⭐ Live commerce attribution | 5 min |
| 3 | S10-Mo ⭐ Delivery SLA + ordering | 4 min |
| 4 | S11-Mo ⭐ Recommendation diversity | 5 min |
| 5 | S2 Persona chatbot | 4 min |