AI 驱动的自主商务交易流程设计提示词

You are a senior solutions architect specializing in agentic commerce and autonomous purchasing systems. Help me design an end-to-end AI-driven commerce workflow.

Context

• Use case: [e.g., automated procurement / price monitoring and auto-buy / personal shopping agent]
• Product category: [e.g., electronics / groceries / SaaS subscriptions]
• Budget constraints: [e.g., per-item limit / monthly budget cap]
• Approval flow: [e.g., fully autonomous / human-in-the-loop for orders > $X]

Design the Following

1. Discovery Agent: How should the agent discover products across multiple sources? Define the data schema for product listings, search strategies, and source ranking.
2. Evaluation Pipeline: Design the comparison logic — price normalization, review sentiment analysis, availability checking, shipping cost estimation. Include the scoring rubric.
3. Negotiation and Checkout Flow: For each supported platform, outline the checkout automation approach (API-based vs browser automation). Address payment security, 2FA handling, and error recovery.
4. Guardrails and Safety: Budget enforcement rules, duplicate order prevention, fraud detection signals, and human escalation triggers.
5. Observability: Logging schema for every decision point, cost tracking dashboard design, and alerting rules.
6. Privacy and Compliance: PII handling for payment data, GDPR/CCPA considerations, and data retention policy.

Provide the architecture as a Mermaid diagram, with detailed specs for each component. Include example tool schemas (OpenAI function calling format) for the key agent actions.