Skip to main content

Module 37: Advanced - Building a Personalized Shopping Agent

Theory​

Introduction​

In this advanced module, we'll build a sophisticated AI agent capable of navigating a simulated e-commerce website to help a user find and purchase a product. This agent will demonstrate how to integrate external web environments and use a combination of tools to perform complex, multi-step tasks.

Agent Architecture​

The personalized shopping agent is a distributed multi-agent system composed of three specialized agents communicating via the Agent-to-Agent (A2A) protocol:

  1. Orchestrator Agent: The main, user-facing agent. It manages the conversation, understands user intent (including multimodal image input), and delegates tasks to the appropriate specialist.
  2. Personalization Agent: A remote agent responsible for managing user preferences. It uses the ADK's state management features to remember information like preferred sizes, colors, and brands across sessions.
  3. Web Agent: A remote agent that acts as an interface to the e-commerce website. It exposes a set of OpenAPI tools for searching products and clicking buttons, abstracting the web environment from the main orchestrator.

Core Components​

  1. Web Environment: The agent interacts with a simulated webshop environment provided by the web_agent_site library. This environment mimics a real e-commerce website, with pages for search, product details, descriptions, features, and reviews. The environment is stateful, meaning it keeps track of the current page and updates it based on the agent's actions.

  2. Tools: The agent is equipped with two primary tools:

    • search(keywords: str): This tool takes a search query from the agent, passes it to the webshop environment, and returns the HTML content of the search results page.
    • click(button_name: str): This tool simulates clicking a button on the current webpage. It takes the name of the button to click (e.g., "Next >", "Description", "Buy Now"), updates the environment's state, and returns the new HTML content.

  3. Prompt Engineering: The agent's instruction prompt is crucial for its success. It defines a state machine-like flow that guides the agent through the shopping process:

    • Initial Inquiry: Ask the user for the product they're looking for.
    • Search Phase: Use the search tool and present the results.
    • Product Exploration: Use the click tool to navigate to product details, descriptions, features, and reviews.
    • Purchase Confirmation: Use the click tool to select options and confirm the purchase.
    • Finalization: Inform the user that the purchase is complete.

    The prompt also includes specific instructions on how to handle the web environment's state, such as using the "< Prev" button to navigate back.

  4. Artifacts: To provide a better user experience, the agent saves the HTML content of the current page as an artifact after each action. This allows the user to see the web page that the agent is interacting with in the ADK's web UI.

By combining these components, we can create a powerful agent that can navigate a web environment, gather information, and interact with a user to complete a complex task.

Key Takeaways​

  • This capstone project integrates many concepts from the course: distributed multi-agent systems (A2A), state management, OpenAPI tools, multimodal input, and artifacts.
  • The architecture separates concerns into a main Orchestrator, a stateful Personalization Agent, and a Web Agent that abstracts the web environment.
  • The agent's instruction is engineered to follow a state machine-like process, guiding it through the complex, multi-step task of navigating a website.
  • Artifacts are used to provide a better user experience by visualizing the agent's interaction with the web environment.
  • Abstraction via OpenAPI: Abstracting the website behind an OpenAPI spec is a superior design because it simplifies the orchestrator's reasoning task. The LLM only needs to know about the search(keywords: str) function, not the complex and messy raw HTML of the website. This improves reliability and makes the system more maintainable, as changes to the website's front-end only require updating the Web Agent's internal logic, not the orchestrator or the OpenAPI contract.
  • Observability via Callbacks: Using a before_tool_callback for logging separates the concern of observability from the agent's business logic. The orchestrator's core instruction remains focused on delegation, while the callback transparently intercepts and logs the action. This makes the system more maintainable, as the monitoring logic can be updated independently of the agent's reasoning.
  • Advantages of Distributed Architecture: This distributed A2A architecture offers significant advantages over a monolithic agent. It allows for independent scalability (the Web Agent can be scaled separately if it's under heavy load), modular maintenance (changes to the website's logic only require updating the Web Agent), and reusability (the Personalization Agent could be reused by other agents in the organization).