Frame the system
Define the audience, job, constraints, states, success criteria, and technical context after discovery and competitive analysis.
Expertise
Product + UX consultancy for AI-made software
GENERATED ≠ VALIDATEDAI helps you ship fast. Not necessarily the right thing.
20+ years of UX and product judgment, turning AI speed into product velocity.
Trusted by:
Select your role
View services and related case studies that most align with your role’s needs.
AI methodology
AI across the workflow. Human judgment at every handoff.
From research and rapid iteration to enterprise-grade production code, I combine design, coding, and AI surgically—using each where it creates the most leverage.
The process starts after discovery and competitive analysis. AI accelerates the first pass and the repetitive work; direct design edits, source-code inspection, and experienced judgment preserve the intent as the product moves from an idea to a maintainable system.
⌘ Current production workflow
From design intent to a live, maintainable system
The tools vary by project, but the operating model stays consistent: establish the problem, generate quickly, refine directly, and keep the final experience editable and owned.
Prompt the first direction
Pair a rough hand-drawn wireframe with visual references and a specific prompt to generate the initial HTML or Figma direction.
Design through a hybrid loop
Iterate with prompts, direct hand edits, and MCP-connected Figma. Move individual components between canvas and code as needed.
Build + inspect the code
Use AI to scaffold HTML, CSS, JavaScript, React, or Python, while continuously reviewing the source for bloat, drift, and over-engineering.
Test in real context
Validate responsive behavior, content density, accessibility, edge cases, consistency, and production feasibility in a working prototype.
Prompt where it is fast
AI is excellent for exploration, scaffolding, variations, and repeatable production. It gets the work moving before polish becomes expensive.
Edit where precision matters
Granular prompting eventually creates latency and collateral changes. Direct Figma and code edits protect design integrity without sacrificing AI’s speed.
Stay close to the source
Constant code inspection catches compounding bloat early. A background in Python, Scheme/Lisp, and data structures also supports practical stack and back-end decisions.
Core production: ChatGPT Codex, Claude, Cursor + Cursor Agent · Independent audits: Gemini, Meta, Grok · Media + assets: CapCut, ElevenLabs, AudioOne · Flexible architecture: Python plus Markdown/JSON/JavaScript/HTML/CSS for dynamic behavior without unnecessary infrastructure.
The same principle, extended into the product itself.
SMA is the adaptive-interface framework I created. Signal → Meaning → Action adds a meaning layer between raw context and software execution. In the workflow, judgment guides what AI should produce. In SMA, that logic becomes part of the architecture so interfaces can adapt to people, goals, and changing situations.
Pricing
Start with one useful decision.
Bring the product and the problem to a focused working call.
Product Assessment
$50 / 30 minutes
We review what is stuck, identify the highest-leverage issue, and align on what should happen next. You receive a concise written recommendation after the call.
Book a 30-minute Assessment · $50Focused execution packages
1, 3, 5, and 10 hour packages
Want to skip straight to execution? New clients who purchase a 1, 3, or 5-hour package receive the 30-minute assessment free, in addition to their paid working hours. The 10-hour package includes a complimentary 60-minute kickoff, for 11 total hours. This one-time onboarding benefit applies to your first package as a new client.