Practice
Research that changes what ships.
I work on products where the difficulty is structural: many roles, long workflows, high consequence, and increasingly, a model somewhere in the middle. The job is to find the questions worth resolving, answer them with the least ceremony possible, and stay close enough to the roadmap that the answer lands.
How I think
Research
Understand people, behaviour, needs, context, and the points where the work actually fails.
Systems
Map dependencies, workflows, tools, ownership, data, and the organizational constraints around them.
AI
Evaluate where intelligent systems create value, where they introduce risk, and how humans should interact with them.
Strategy
Translate evidence into priorities, product direction, experience principles, and decisions.
Applied AI toolkit
Across the AI product lifecycle.
I work across the AI product lifecycle—from research and evaluation to prototyping, content generation, and workflow design.
ChatGPT · Claude · Gemini Enterprise · Microsoft Copilot · Copilot Studio
Research & evaluation
AI experience research · test sets · output evaluation · guardrails · human-in-the-loop systems
Build & prototype
Agents · custom instructions · AI-assisted research tools · websites · notebooks · rapid prototypes
Multimodal creation
Image generation · video generation · visual concept development
Workflow design
AI-assisted workflows · orchestration · automation exploration
Methods I lean on
- Systems and workflow mappingTracing a task across every tool, role, and handoff it actually touches.
- Contextual inquiryWatching the work in its own environment, including the workarounds.
- Longitudinal diary studiesFor workflows whose failures only appear over a full cycle.
- Evaluative studiesStructured comparison of concrete alternatives, not preference collection.
- Measurement designDefining quality so two teams can argue about the same thing.
Current areas of focus
- AI opportunity identificationWhere intelligence changes the shape of a workflow, not just its speed.
- Human-in-the-loop workflowsWhere judgement belongs, and what it costs to exercise it.
- AI evaluationShared definitions of quality agreed before the launch debate begins.
- Research operationsMaking evidence cheap enough that teams actually use it.
- Enterprise systemsDepth for experts without abandoning the people arriving new.