AI product definition.
Clarify whether AI should be the product, a workflow layer, an assistant, or simply not there yet.
Use-case selectionUser valueMVP scopeAI vs non-AI boundaries
AI product strategy
Make AI useful, trusted and buildable.
I help founders and teams decide where AI belongs in a product, how the experience should behave, what should stay human-controlled, and how to turn AI ambition into workflows people can actually use.
Where I help
Product discipline for AI work.
Assistant experience design.
Design assistants that understand context, ask for clarification, recover from mistakes and feel useful in real workflows.
Conversation flowsMemory modelClarifying questionsRecovery paths
Human-in-the-loop systems.
Decide where users should review, approve, correct or override AI output before anything important happens.
Review statesApproval flowsConfidence cuesAudit trails
AI workflow architecture.
Turn vague AI ambition into structured workflows, roles, data inputs, outputs, permissions and operational controls.
Context captureWorkflow triggersStructured outputsOperational safeguards
Principles
Trust is the product.
AI should improve the product outcome, not decorate the pitch.
Users need to understand what the system knows, what it guessed and what they can change.
High-stakes actions need review, approval and recovery paths.
Memory is a product decision, not only a technical feature.
The best AI products feel bounded, useful and correctable.
Related work
Next step
