AI 產品推薦

Human → AI

AI 分析顧客的瀏覽與購買紀錄，產生個人化推薦。

5 個節點 · 5 條連接retail

agenthumandb

視覺化

收集顧客資料db

擷取瀏覽紀錄、購買紀錄與偏好訊號。

↓sequential→ 分析偏好

分析偏好agent

AI 識別模式、細分顧客並對產品親和度進行評分。

↓sequential→ 產生推薦清單

產生推薦清單system

產出附有信心分數的產品排名清單。

↓sequential→ 商品人員審查

商品人員審查human

人工審查推薦內容是否符合品牌定位與庫存狀況。

↓conditional→ 推送至門市

↓fallback→ 分析偏好

推送至門市api

將核准的推薦部署至產品頁面與電子郵件行銷活動。

在編輯器開啟

uc-product-recommendation.osop.yaml

osop_version: "1.0"
id: "ai-product-recommendation"
name:"AI 產品推薦"
description:"AI 分析顧客的瀏覽與購買紀錄，產生個人化推薦。"

nodes:
  - id: "browse_data"
    type: "db"
    subtype: "query"
    name: "收集顧客資料"
    description: "擷取瀏覽紀錄、購買紀錄與偏好訊號。"

  - id: "ai_analyze"
    type: "agent"
    subtype: "llm"
    name: "分析偏好"
    description: "AI 識別模式、細分顧客並對產品親和度進行評分。"
    security:
      risk_level: "low"

  - id: "generate_recs"
    type: "system"
    name: "產生推薦清單"
    description: "產出附有信心分數的產品排名清單。"

  - id: "merchandiser_review"
    type: "human"
    subtype: "review"
    name: "商品人員審查"
    description: "人工審查推薦內容是否符合品牌定位與庫存狀況。"
    security:
      approval_gate: true

  - id: "push_storefront"
    type: "api"
    subtype: "rest"
    name: "推送至門市"
    description: "將核准的推薦部署至產品頁面與電子郵件行銷活動。"

edges:
  - from: "browse_data"
    to: "ai_analyze"
    mode: "sequential"
  - from: "ai_analyze"
    to: "generate_recs"
    mode: "sequential"
  - from: "generate_recs"
    to: "merchandiser_review"
    mode: "sequential"
  - from: "merchandiser_review"
    to: "push_storefront"
    mode: "conditional"
    when: "review.approved == true"
  - from: "merchandiser_review"
    to: "ai_analyze"
    mode: "fallback"
    label: "Adjust recommendation criteria"