Evan Cannarozzi

Human First, AI Faster: AI Personas That Actually Help Product Design

Design within teams is hard: constraints, shifting scope, busy stakeholders, and a calendar that does not care about your ideal process. In this workshop, Evan takes a “real-world design operating system” and gives a practical method for a human-first, AI-accelerated feedback loop using AI personas. You will start with real human input, then build a lightweight synthetic user based on that context, run live design feedback against it, and compare the synthetic output to the human response to spot where it aligns, where it diverges, and where it is just confidently wrong. The goal is not to replace research. It is building a faster decision-making workflow that helps small or stretched teams test, refine and communicate clearly with PMs and stakeholders. Practical, realistic and intentionally un-hypey. Augment, do not abdicate. What you’ll walk away with: - A step-by-step synthetic user workflow to capture real stakeholder input, turn it into an agent, and use it to speed up iteration without pretending it is “real users.” - How to configure an AI persona through better setup and prompts: define the right context and boundaries, then generate critique that flags clarity vs friction (less garbage, more signal). - A simple validation loop: how to compare AI feedback to human feedback, spot hallucinations and overconfidence, and use the overlap to prioritise what to fix next. - Faster collaboration with PMs and reduce reruns: How to use the loop to show progress, iterate live, and communicate decisions in outcome language that builds trust across product, design and engineering.