I come from UX and product design. Building lets me follow an idea through the details: how it feels, how it behaves, where it breaks, and what someone can actually do with it.

Through Affordance Design Studio, I bring together design, product thinking, hands-on development and education. My work moves between AI experiences, tools for builders, creative software, and workshops that help people turn an idea into something they can test.

A recurring question behind my projects:

When software can do more on our behalf, how do we make intent, judgement and control easier to preserve?

That question shows up in an agent's permissions, a project's memory, a visual editing tool, and the way I teach people to build.

memory-map packages session snapshots, context recall and memory cleanup into reusable skills. It addresses a familiar frustration: opening a new conversation and having to reconstruct the project before making progress.

TRACE connects a project's original intent to decisions made during development. It proposes updates to requirements, assumptions and open questions, with a review step before those changes become part of the project record.

HAL 9000 explores an AI assistant that combines voice, vision, memory and computer tools. It brings a local application, a browser control panel and MCP integration into one workspace.

VibeLens is a browser extension for inspecting and adjusting CSS visually, with changes written back to source. It brings interface refinement closer to the designer's immediate feedback loop.

Project Workflow Protocol collects planning practices, verification steps, templates and AI collaboration rules. It makes scope, authority and completion criteria explicit.

Signal monitors usage across Claude Code and Codex CLI through a terminal interface and a visual web view. There is also a pixel-art crab. Useful tools can have some character.

Each repository has its own setup, limitations and project status. Start with its README before running anything.

At Affordance, my interests meet in three areas:

• Designing AI experiences: how people understand a system, direct it, recover from errors and decide when to trust it.
• Building products and tools: turning design decisions into working behaviour, then learning from what happens.
• Developing people's capabilities: helping designers, students and teams build with AI while exercising their own judgement.

My teaching includes Vibe to Viable, workshops and build programmes that move through problem definition, prototyping, implementation and critique. I like a session to leave people with something they made, something they tested, and a clearer sense of what to do next.

Explore Vibe Coding School · Build with Claude

1. Make the intent clear. Describe the problem, the person and the outcome before choosing the implementation.
2. Build something inspectable. A working interaction makes a discussion more useful.
3. Keep human decisions visible. Let agents prepare and propose; make consequential approvals explicit.
4. Check the result. Distinguish a generated artifact, a passing test and a useful experience.
5. Leave context for the next session. Record what changed, why it changed and what remains unresolved.

I work across TypeScript, JavaScript and Python, with interests spanning browser interfaces, agent tools, MCP integrations, persistent context, creative interaction and workflow automation.

I am particularly interested in:

• Agents that can explain what they are about to do and ask at the right moment.
• Design tools that shorten the distance between an idea and a working interface.
• Project memory that stays useful as the work changes.
• Learning environments where people build, question and revise.
• Small utilities that remove a recurring frustration from the day.

For product and design collaborations, AI experience work, workshops or teaching conversations, find me through Affordance or LinkedIn.

For a specific project, an issue with clear context, a reproducible bug, or a thoughtful improvement is a great place to start.