From prompts to Practice
Embedded Files
Building AI Agents for UX research, Design, and Design systems
I am exploring how tools like ChatGPT, Microsoft Copilot, and Claude can help simplify and improve my everyday UX activities across research, product design, and design-system development.
My focus is on creating reusable AI agents that reduce repetitive work, bring consistency to my processes, and allow me to spend more time on problem-solving, strategic thinking, and creating meaningful user experiences. Rather than writing prompts from scratch for every task, I am developing a structured blueprint for each agent, defining its role, objectives, inputs, methodology, workflow, expected outputs, and quality standards.
My Responsibilities
As an Experience Strategist, my responsibility is to design, guide, and continuously improve these AI-assisted workflows.
● Identify opportunities: Find repetitive, time-consuming, or inconsistent UX activities suitable for agent support.
● Design agent blueprints: Define roles, instructions, inputs, methodologies, and outputs.
● Establish standards: Ensure agents follow UX principles, research practices, accessibility requirements, and design-system conventions.
● Validate outputs: Review generated results for accuracy, relevance, evidence, and usability.
● Iterate and optimise: Test agents against real scenarios, identify gaps, and refine their instructions.
● Enable adoption: Document workflows and help teams use agents effectively and responsibly.
AI supports the execution; professional judgement remains essential for interpreting evidence, making decisions, and ensuring user-centred outcomes.
Agents
Blueprint
Every agent follows a common blueprint:
● Role: Expertise and perspective the agent should adopt.
● Objective: The task or problem it needs to address.
● Context: Product, user, business, and project information.
● Inputs: Information required to execute the task.
● Methodology: Frameworks, principles, and processes to follow.
● Workflow: Step-by-step execution instructions.
● Output: Expected deliverables and their format.
● Validation: Quality, accuracy, accessibility, and evidence checks.
● Iteration: Feedback-driven refinement and continuous improvement.
Key Value Adds
Reduce repetitive manual work and accelerate the delivery of UX research, design, and documentation activities.
Establish reusable workflows and standardised outputs that improve the consistency and reliability of UX deliverables.
Free up capacity for deeper research, better product decisions, experience innovation, and design-system evolution.
"Guides, prompts, instructions, skills, connectors and vibe-coded inspiration for AI product design."
Google Sites
Report abuse