Product design workflow

Workflow Modes

One system, different speeds.

Same workflow, different fidelity — never different principles.

"Full flow" ensures the end-to-end user experience is complete and unbroken, even in fast or lightweight projects. "Full workflow" ensures the design process itself is intentional, systemized, and learnable — required for platform-level or strategic initiatives.

Principles

Four commitments that don't compress.

This workflow reflects our 2026 direction: global collaboration, AI-native product strategy, system-based thinking, and async execution at scale.

1. Async-First Collaboration

• Work happens in documents first, not meetings.
• Meetings are for decision-making, not information transfer.
• Default tools: Notion, FigJam, Figma, Loom, Slack.

2. AI-Native By Default

• AI is embedded across the workflow to accelerate research, synthesis, exploration, validation, and learning.
• AI never replaces judgment — it compresses cycles.
• Designers own intent, trade-offs, and final decisions.

3. System-Based, Not Screen-Based

• We design experience systems, lifecycle models, behavior rules, and reusable components.
• Screens are outputs, not the goal.

4. Global Team, Low-Misunderstanding Communication

• Explicit writing over implicit assumptions.
• Clear ownership and decision logs.
• Artifacts over opinions.

Stage 1 · Discovery

Shared understanding before design.

Goal: Align on the real problem before touching solutions.

Tools

• FigJam — collaborative framing
• Notion — async review & context
• AI Research Assistant (GPT / Gemini) — synthesis
• Loom — async context
• Slack Threads — Q&A

Activities

• JTBD framing
• Task universe mapping
• Persona & lifecycle analysis
• Research input gathering
• Competitive teardown (AI-assisted clustering)
• Problem articulation

Deliverables

• Problem Framing Canvas
• Research Inputs Wall
• AI Competitive Summary
• Stakeholder Alignment Note
• North Star Definition

AI speeds up research, async tools reduce meetings, FigJam drives alignment.

Stage 2 · Definition

From insights to direction.

Goal: Turn ambiguity into design intent and boundaries.

Tools

• FigJam — IA & concept models
• AI Synthesis (GPT / Gemini) — pattern detection
• Notion — scope & definition docs
• Loom — async walkthroughs

Activities

• Insight consolidation
• Task prioritization (value × frequency)
• Concept model creation (e.g. Lifecycle Engine)
• IA drafting
• Data dependency identification
• Behavior rule hypotheses

Deliverables

• Concept Model v1
• IA v1
• Behavior Rule Draft
• Scope Definition

AI helps compress insights → patterns faster.

Stage 3 · Exploration

System primitives, not pixels.

Goal: Explore broadly using system primitives, not pixels.

Tools

• Figma Make — structured exploration
• Design System Tokens
• AI Design Assistant (GPT / Gemini) — layout suggestions, state variations
• FigJam — async feedback on early options

Activities

• System-first wireframing
• Layout explorations
• Multi-lifecycle variants
• Cross-channel consistency checks
• Low-fi prototypes

Deliverables

• Wireframes
• Component proposals
• Interaction sketches

AI quickly generates multiple variations — designers curate and refine.

Stage 4 · Blueprinting

Remove ambiguity before execution.

Goal: Remove ambiguity before hi-fi execution.

Tools

• FigJam — flows & logic
• AI Flow Generator (GPT / Gemini) — generate drafts
• Notion — acceptance criteria
• Slack Threads — Eng/PM Q&A

Activities

• End-to-end flow definition
• Personalization & behavior rules
• Data requirement mapping
• Feasibility alignment
• Edge case clarification

Deliverables

• Experience Blueprints
• Behavior Rule Spec
• Data Map
• Acceptance Criteria

Blueprinting eliminates 80% of preventable rework.

Stage 5 · Execution

Polished, systemized UI.

Goal: Deliver high-quality, reusable UI with confidence.

Tools

• Figma Make — hi-fi systemized UI (DS-bounded)
• Design System — tokens, primitives, patterns
• Loom — async walk-throughs for PM/Eng
• Slack — design reviews

Activities

• Hi-fi UI design
• Responsive layouts
• Micro-interactions
• DS updates
• Review cycles

Deliverables

• Hi-Fi Screens
• Prototypes
• Redlines
• Component variants

Designer owns quality — AI accelerates precision, not judgment.

Stage 6 · Validation

De-risk before shipping.

Goal: Validate usability, clarity, and value.

Tools

• FigJam — test planning & synthesis
• Maze — unmoderated usability testing to quickly validate task completion, clarity, and friction at scale
• Moderated testing — facilitated sessions to observe user behavior and probe mental models, language understanding, and perceived value
• AI Interview Summaries — speed up synthesis
• Loom — async review of findings

Activities

• Test planning
• Usability testing (unmoderated + moderated)
• Insight synthesis
• Issue prioritization

Deliverables

• Findings report
• Insight clusters
• Priority matrix
• Iteration plan

AI summarization reduces analysis time from days to hours.

Stage 7 · Delivery

Zero confusion at handoff.

Goal: Enable engineering to ship without confusion.

Tools

• Figma Make → Spec mode
• Notion → Handoff packet
• Loom → Async walkthrough
• Slack threads → Q&A log

Deliverables

• Final UI
• Behavior rules
• Analytics events
• Edge states
• QA checklist

Expect "zero confusion" at handoff — engineering ships without a single clarification meeting.

Stage 8 · Post-Launch

Close the loop.

Goal: Turn releases into system improvements.

Tools

• AI Data Insights (GPT / Gemini)
• FigJam — insights wall
• Notion — decision log updates
• Loom — async review with PM

Activities

• Post-launch analysis
• Merchant feedback synthesis
• DS updates
• V2 opportunity discovery

Deliverables

• Post-launch insights
• Updated assumptions
• Component graduation proposals
• V2 opportunity board

We design a system, not just features.

🔁 Learning Loop

• Concept Model → updated
• Behavior Rules → refined
• Components → promoted or deprecated