Lesson 2: Understanding AI Agents as Users

Lesson 2: Understanding AI Agents as Users

Introduction: AI Agents as the New Digital Users

For decades, digital experiences were designed exclusively for human users. Websites, applications, and systems focused on human-friendly navigation, intuitive interfaces, and user experience (UX) best practices. But as AI agents—such as virtual assistants, search crawlers, and automation bots—become more prevalent, digital platforms must also cater to AI users.

Just like human users have personas, behaviors, and needs, AI agents have their own interaction models, workflows, and constraints. Understanding AI agents as users is crucial to optimizing AI-driven automation, searchability, personalization, and decision-making.

Defining AI Agent Personas

What is an AI Agent Persona?

Just as UX designers create user personas to understand different types of human users, AI agent personas help define how different AI systems interact with digital platforms. These personas categorize AI agents based on:

  • Their core function (e.g., searching, decision-making, automation).

  • Their input sources (structured data, APIs, NLP, logs, etc.).

  • Their goal-oriented behavior (retrieving, analyzing, executing tasks).

Creating AI agent personas helps businesses optimize digital platforms for AI-driven interactions—ensuring that AI systems can efficiently navigate, retrieve, and act on data.

Types of AI Agents

AI agents fall into four major categories, each with unique behaviors, interaction models, and optimization needs.

1. Personal Assistants (Conversational AI)

  • Examples: Siri, Google Assistant, Alexa, ChatGPT-powered assistants.

  • Primary Role: Understanding natural language queries and retrieving relevant information.

  • Interaction Mode: Voice or text-based interactions with API-backed data sources.

  • Key Optimization Needs:
    ✅ Structured data for clear information retrieval.
    ✅ Conversational UX design for seamless responses.
    ✅ API integrations to fetch live updates and actions.

📌 Example Scenario: A user asks, "What’s the best-rated sushi restaurant nearby?" – The assistant queries Google Places API, retrieves structured data, and ranks results accordingly.

2. Search & Indexing Agents

  • Examples: Googlebot, Bingbot, AI-powered product recommendation engines.

  • Primary Role: Crawling and indexing structured data to provide search results.

  • Interaction Mode: Parsing HTML, schema.org metadata, and API-driven content.

  • Key Optimization Needs:
    ✅ Machine-readable content (JSON-LD, Microdata, RDFa).
    ✅ SEO and structured metadata to enhance indexing.
    ✅ Fast-loading, query-efficient API endpoints.

📌 Example Scenario: Googlebot crawls a webpage but fails to detect structured data—resulting in lower search rankings and AI misinterpretation.

3. Decision-Making AI Agents

  • Examples: Fraud detection AI, credit risk assessment bots, recommendation engines.

  • Primary Role: Analyzing data, patterns, and risks to make predictions and automate decisions.

  • Interaction Mode: Consumes structured and unstructured data from multiple sources.

  • Key Optimization Needs:
    ✅ High-quality, real-time data feeds for accurate predictions.
    ✅ Transparent AI logic with bias detection safeguards.
    ✅ Ability to integrate external APIs for context-aware decisions.

📌 Example Scenario: A credit scoring AI analyzes a user’s transaction history to approve or deny a loan request in milliseconds.

4. Task Automation Agents

  • Examples: AI-powered workflow bots, RPA (Robotic Process Automation), customer support chatbots.

  • Primary Role: Executing automated tasks based on predefined triggers.

  • Interaction Mode: API integrations, rule-based workflows, and event-driven execution.

  • Key Optimization Needs:
    ✅ Efficient API automation for fast execution.
    ✅ Workflow orchestration to manage multi-step processes.
    ✅ Scalability to handle thousands of automated actions.

📌 Example Scenario: An AI-driven support chatbot automatically refunds a customer without requiring human intervention.

How AI Agents Perceive and Process Information

1. How AI Agents "See" Digital Content

Unlike human users who visually scan websites, AI agents interpret content through structured data, APIs, and machine-readable formats.

  • Human Perception: Reads text, views images, processes UI elements.

  • AI Perception: Extracts structured data (JSON, XML, schema.org).

  • Optimization Tip: Use schema markup and well-documented APIs to improve AI accessibility.

📌 Example:
❌ A product page without structured data → AI agent fails to categorize it properly.
✅ A product page with JSON-LD markup → AI agent correctly identifies product details (price, brand, availability).

2. How AI Agents Process Information

AI agents process information through structured workflows, unlike humans who use intuition and emotions.

AI Processing Steps:
1️⃣ Input Parsing: Extracting structured data, metadata, or API responses.
2️⃣ Data Processing & Analysis: Running machine learning models, rule-based logic, or decision trees.
3️⃣ Action Execution: Sending API requests, generating content, or making predictions.
4️⃣ Learning & Adaptation: Refining future interactions based on user feedback or retrained models.

📌 Example Scenario: A search engine AI crawling an e-commerce site processes:
🔹 Title Tags & Descriptions → Helps AI categorize products.
🔹 Schema.org Markup → Enables feature-rich search snippets.
🔹 Internal Linking Structures → Improves AI's understanding of product relationships.

Key Takeaways:

✅ AI agents are the new digital users—they need structured, machine-readable environments.
✅ Different AI agents have different optimization needs—personal assistants need NLP, search agents need structured data, decision-making AI needs real-time inputs, and automation agents need APIs.
✅ AX (AI Agent Experience) requires a shift from human-centric to AI-friendly design—leveraging APIs, structured metadata, and agent-first workflows.

🚀 Next Lesson: Agentic Experience Mapping (AX Maps) – How to Design Workflows for AI Agents!

Would you like a real-world case study of AX design in an upcoming lesson? Let us know! 👇

AI Agent Experience (AX)Francesca Tabor