- From: Gaowei Chang <chgaowei@gmail.com>
- Date: Tue, 15 Jul 2025 13:03:30 +0800
- To: public-agentprotocol@w3.org
- Message-ID: <CAGJoCKysyps0gja6DB4+DUonvwV7PmF8S1GiUgHt9GTWOJ_iZA@mail.gmail.com>
hi, all: How agents will collaborate in the future is a crucial premise for designing agent protocols. I’ve added a new document that outlines a possible use case based on my perspective. I hope we can discuss it together, and I’d love to hear your thoughts on how you envision agent collaboration evolving in the future. https://github.com/w3c-cg/ai-agent-protocol/blob/main/use_case.html The github issue is https://github.com/w3c-cg/ai-agent-protocol/issues/15 The first use case: *Scenario Overview:* Dixon is planning a vacation to Paris and needs to book a hotel that matches his preferences. He decides to let his Personal Agent handle this task, demonstrating the complete workflow of modern agent collaboration. *Step 1 - Requirements Analysis and Search:* Dixon's Personal Agent, based on previously authorized access to personal data (including hotel preferences: 4-star or above, close to city center, with gym; vacation dates: March 15-22; budget range: 200-400 euros per night), uses the agent protocol to send a structured search request to the Search Agent. The Search Agent performs matching within its maintained global hotel agent network and quickly returns information for 100 qualifying hotel Service Agents. *Step 2 - Information Collection and Filtering:* The Personal Agent uses Dixon's identity credentials to concurrently access these hotel Service Agents, requesting detailed information (real-time room rates, available room types, specific locations, facilities and services, user reviews, etc.). Each hotel Service Agent, recognizing potential customer access, proactively provides detailed service information and promotional offers. After collecting all information, the Personal Agent applies Dixon's personal preference algorithm to filter 8 best-matching hotel options from the 100 choices, ranking them by match score before pushing them to Dixon. *Step 3 - Booking and Payment:* After reviewing the Personal Agent's recommended options, Dixon selects the second-ranked boutique hotel (because the top-ranked hotel had noise issues mentioned in reviews). The Personal Agent immediately establishes a booking session with that hotel Service Agent, completing room reservation and booking confirmation. However, due to security policies, the Personal Agent lacks payment authorization. After Dixon receives the successful booking notification, he completes payment through a secure payment interface. Once payment is completed, the Personal Agent automatically notifies the hotel Service Agent of the payment status and saves the complete booking information (confirmation number, check-in details, hotel contact information) to Dixon's personal travel profile. *Collaboration Value:* Throughout this process, the Personal Agent serves as Dixon's digital representative, the Search Agent provides agent discovery services, and the hotel Service Agent provides professional services. All three achieve seamless collaboration through standardized agent protocols, significantly improving booking efficiency and user experience. *Requirements:* Agent Identity Authentication <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#agent-identity-authentication> , Agent Discovery <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#agent-discovery>, Agent Description <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#agent-description> , Agent Information Interaction <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#agent-interaction> , Agent Transactions <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#agent-transactions> , Privacy Protection <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#privacy-protection> , Identity Interoperability <https://w3c-cg.github.io/ai-agent-protocol/use_case.html#identity-interoperability> Looking forward to the discussion! Gaowei Chang
Received on Tuesday, 15 July 2025 05:03:49 UTC