top of page

My Product Design Process with AI

Updated: 06/24/2026

I use AI tools to speed up routine tasks so I can spend more time on product strategy, complex data architecture, and UX problem-solving.

 

Here is a sample workflow for how I design products, combining human judgment and AI efficiency.

 
AdobeStock_303329613.jpeg

01

Stakeholder Alignment

My Role

Leading kickoff sessions to cross-functionally align product management, engineering leads, and business stakeholders. I use audio recording and LLMs to instantly parse stakeholder intent and convert complex infrastructure realities, backend data limitations, and system constraints into a production-ready product

Tools

Audio recording, Gemini, Claude

Deliverables

Core product vision document, technical and engineering limitation guardrails, quantified success metrics, and user-behavior assumption maps.

competitive analysis.avif

02

Market Research

My Role

Conducting competitive landscape audits across direct and indirect market players. I use LLMs to synthesize feature-parity data, identify market opportunities, and filter out generic AI biases.

Tools

Gemini, Claude

Deliverables

Strategic Market Positioning Map, feature-parity benchmark matrices, and a prioritized unique value proposition brief.

user journeys.png

03

Personas, journey & flow

My Role

Pulling raw qualitative and behavioral data such as support tickets, sales call transcripts, and user feedback directly into LLMs to build data-driven user personas. From there, I map end-to-end journeys to pinpoint where users face cognitive load or interaction friction.

Tools

Claude, FigJam, Gemini, ChatGPT.

Deliverables

Personas, user journey maps, connected primary user flows, and a prioritized user pain-point log.

panorama-shot-frontend-developer-team-26

04

Mid-fi wireframes

My Role

Partnering with backend engineers to map out data requirements and system parameters. I use Claude Design to quickly explore structural directions, then move the strongest layout into Figma for refinement. I skip lo-fi sketching and work with production-intent content from the start.

Tools

Claude Design, Figma, Figma Make

Deliverables

Component-ready mid-fi wireframes, cross-functionally agreed structural layout direction, and production-intent data schema requirements.

user-testing_edited.jpg

05

Usability testing

My Role

Running validation cycles with 3 to 5 users or internal teammates. Before sessions, I use AI attention analysis to predict where users will focus and catch layout issues early. I evaluate interaction patterns against UX heuristics and use LLMs to synthesize test transcripts into optimization fixes.

Tools

Deliverables

AI-synthesized usability test reports, pre-test attention heatmaps, verified heuristic friction points, microcopy fixes, and interaction adjustment lo

AI infrastructure use case thumbnail.png

06

Hi-fi mockups

My Role

For new products: I use Claude Design to quickly explore visual directions, then build a scalable variable system in Figma.

For existing products: I import the live design system into Claude Design so output is validated against real tokens before I touch it. I then adjust properties in Figma, bulk-edit variations in Codex or ClaudeCode while enforcing token constraints, and return to Figma to prevent system drift.

Tools

Figma, Figma Make, Claude

Deliverables

High-fidelity Figma files, variable-driven component libraries, and finalized UI copy.

Screenshot 2026-05-24 at 9.04.20 PM.png

07

Engineering handoff

My Role

Walking engineers through the Figma file live. I use Claude Design's two-way Claude Code sync so engineers pull the exact design state into their terminal without rebuilding from screenshots. I write spec notes for anything Figma cannot show and keep them available throughout the build.

Tools

Figma dev mode, Claude, ChatGPT

Deliverables

Figma dev mode, Claude Design, Claude Code, Claude or Gemini

AdobeStock_236135448.jpeg

08

Launch & Learn

My Role

Tracking how real users interact with the live product. I pipe clickstream data and session maps from Microsoft Clarity into LLMs to cluster friction points, trace heatmaps, and surface ignored paths.

Tools

Microsoft Clarity, Claude or Gemini

Deliverables

Tracking how real users interact with the live product. I pipe clickstream data and session maps from Microsoft Clarity into LLMs to cluster friction points, trace heatmaps, and surface ignored paths.

bottom of page