Skip to main content
Brian Williams.

AI as a product design accelerator

Faster exploration. Better-informed decisions.

AI helps me move from problem to tangible prototype faster. It does not replace user context, design judgment, or accountability for the final experience.

01 / DISCOVERY

ChatGPT + Gemini

Synthesize supplied research, analyze requirements, generate interview questions, and explore edge cases.

02 / DESIGN

AI-assisted exploration

Explore content structures, alternative workflows, and scenarios to consider before choosing a direction.

03 / PROTOTYPING

Coded prototypes + coding assistants

Build interactive prototypes and implementation experiments; examine responsive behavior and interaction feasibility.

04 / COMMUNICATION

Higgsfield

Explore concept visualization and visual storytelling for communicating an idea.

Human responsibility

AI changed my velocity — not my responsibility.

I remain accountable for problem framing, user empathy, prioritization, product judgment, tradeoffs, ethics, and final design decisions. AI output is a prompt for review, not evidence of user needs or product impact. I do not use this portfolio to claim AI processed confidential client material.

Independent product exploration · Not client work

Smart Queue Exploration

Users working through a high-volume queue may need help identifying which items deserve attention first without giving up control of the decision. This fictional exploration makes that product question tangible.

Frame → explore → design → code

  1. 01 / Frame

    How might an AI-assisted system highlight potentially important work while keeping the human operator in control?

  2. 02 / Explore

    Use ChatGPT or Gemini to generate edge cases, challenge assumptions, compare ranking explanations, and identify failure modes—not as user research.

  3. 03 / Design

    Make the recommendation, rationale, uncertainty, and available human actions visible together.

  4. 04 / Code

    Build an interactive queue prototype and inspect its behavior directly.

Product judgment / decision changed

Initial idea

AI automatically reorders the queue.

After review

AI recommends priority; it does not silently reorder work.

Operators need to understand and retain control over consequential prioritization.

Interaction states to inspect

Uncertain recommendation · user disagreement · stale signal · missing explanation · incomplete data. The prototype is an interaction demonstration and self-review, not validated product research.

AI assisted: option and edge-case generation, content alternatives, and prototype coding. I owned: framing, interaction model, risk choices, hierarchy, tradeoffs, and final product judgment.

Open live prototype

Working loop

Frame the question → explore options → make an interaction tangible → review behavior and constraints → refine the design decision.