Design is reinventing itself…
…and designers have to get this right. This is my playbook for AI-enhanced design execution that complements my human-defined product vision.
AI design principles
The fundamentals won't change, but everything else will.
AI is a force multiplier. Well-crafted agent boundaries help produce immense value.
Winning teams understanding AI's constraints. LLMs hallucinate, reflect bias, and return non-deterministic output. Safeguard against these with well-defined skills and agent workflows. Use AI as an extension of critical thought, not its replacement.
Knowing what to build is more important than ever. Good design solves the right problems.
Craft the right kind of speed
Speed without understanding creates no value. Winning teams prioritize domain knowledge while using AI to accelerate execution.
My current stack
Most of my day is spent in these tools, but I'm adaptable.
Notion MCP
For maintaining a scalable agent-facing context bank
Claude Code
For prototyping in-code, and orchestrating build agents
Figma Agent + MCP
For canvas iterations and fine-tuning
Linear
For project management and build orchestration
Claude Design
For novel UI ideation and system-aligned prototypes
Todoist MCP
For personal task management
Favorite AI techniques
I'm exploring how to build effectively and responsibly in the AI era.
Prototyping in code
With some tuning, AI lets me shrink the gap between concept and production, leading to more productive cross-functional conversations.
Models are no substitute for design critique, but can be useful as a devil's advocate when given sufficient project context.
Agents can churn out established patterns (ie, empty states) while I focus on the more important product questions.
No more time spent reviewing certain data; using agents to assess when the data shape is compatible with LLM pattern-matching.