Lab
Build with AI
AI is moving quickly, which makes first principles more important, not less. I use prototypes to test when conversation helps, when structure helps more, and how explainability changes trust.
Strategy
Useful AI reduces effort, not adds to it
My recurring belief: useful AI should reduce effort, reduce ambiguity, and fit the workflow instead of demanding that people adapt themselves to the tool.
I explore guided dashboards, service conversations, retrieval quality, embedded assistance, and explainable systems. The lab is where theory meets practice: small experiments that sharpen how I think about better product experiences.
Execution
Live prototypes
These public experiments are the practical side of that point of view.
Bon Courage Travel Guide
AI-assisted travel planning
Explores how structured guidance and conversational design can reduce planning friction and help travelers make confident decisions.
AI Dashboard Strategy
AI intention in dashboards
Tests how structured insight modules can help people decide faster without making chat the default interface.
Rate Advisor
Useful AI for everyday decisions
Explores how comparison logic and explanation can reduce decision fatigue in a high-friction consumer choice.
Airline Support Chat
Conversational design
Tests how tone, task framing, and guided prompts can make support interactions feel clearer under pressure.
Toolkit
Skills & agents I enjoy using
As a UX strategist, I don't need to reinvent the wheel on AI — people around the world are already building great tools. This is my curated list, plus a few of my own experiments. There's no value starting from scratch when you can build on a strong base.
Skills I use
A few of my favorites, built by their creators — I didn't build these, so here they are as a simple list, linked straight to the file.
- /find-animation-opportunities — Emil Kowalski, Linear
- /better-interface — Jakub Krehel, Interfere
- /component-handoff — Mauro Berteri, Itti Digital
- /ui-state-expander — Linda Ojo, Paystack
My UX team review
Imagine tapping a brain trust of detail-oriented reviewers before a design ever reaches a critique or a paper-cut backlog sprint. This is that team — they assess, grade, and give feedback before it gets in front of real people.
I built it as one packaged Claude Code skill, not four separate agents — one job, one scorecard. It reviews a Figma file, live code, or a public site, and every reviewer is scoped to exactly one cited source, never invented standards.
Also worth a look
Claude Code is still my daily driver. I also reach for OpenCode, and use Agentation to keep feedback loops tight.
Want to talk through how this kind of thinking could apply to your product?