AI Strategist

Pioneering intelligent systems that understand human context and intent

  • Agentic UX Framework
  • AIPET Methodology
  • AI Product Strategy

UX Architect

Crafting human-centered experiences through evidence-based design

  • Human-Centered Design
  • User Research
  • Design Systems

Pioneering Agentic
User Experience

Designing proactive, trustworthy AI companions that foster independence, dignity, and connection, placing user privacy and control at the forefront.

Core Competencies

AI & Agentic UX

  • AI agent workflow
  • Agentic UX
  • Human-Centered AI
  • AI Trend Analysis & Foresight

Privacy & Ethics

  • Privacy-first Design
  • AI Ethics & Governance
  • GDPR & PIPEDA Compliance
  • Data Governance Strategy

Design & Strategy

  • UX/Design Leadership
  • AI Product Strategy & Roadmapping
  • Service Design Strategy
  • Startup Consulting & Mentoring

Agent Governance Column

Original analysis on AI literacy, dark patterns, and privacy-UX. Writing weekly for 1,000+ subscribers. Was a “Top 25 Rising” newsletter in 2025 and is a “Top 9 Rising in Design domain” newsletter in 2026 Q1.

AGENTIC UX · CONTENT ANALYSIS v2 2026 Substack: Cross-Analysis of 95 Articles (Revised) Method: Term-weight calibration (T broad-terms ×0.4–0.5) + multi-axis primary (top-2) + modern Privacy / Experience vocabulary added Stats: 131 axis tags total (incl. secondary) | 95 articles | vs v1: P 8→23, E 1→8, T 55→54 Article Clusters AIPET Primary Axes (multi-axis) Citation Types Society 42 tags Business 34 tags UX Design 31 tags Education 10 tags Technology 11 tags Healthcare 2 tags Other 1 tag T Trust / Governance 54 tags (41%) I Interaction 37 tags (28%) P Privacy 23 tags (18%) A Agency 9 tags (7%) E Experience 8 tags (6%) Industry 77 tags (59%) Policy 7 tags (5%) Academic 25 tags (19%) Mixed 22 tags (17%) KEY REVISIONS P 8→23 (modern Privacy vocab added) | E 1→8 (friction / breakpoints / continuity added) | A 3→9 (concrete proxy-action terms added) | T 55→54 (still primary after broad-term calibration) Multi-axis primary: each article takes its top-scoring dimension; runner-up included if ≥5 and ≥50% of primary score. Hence 131 tags > 95 articles.

Column Highlights

— Wiki-as-Interface

How AI/UX Writing’s CoT Orchestrates Multiple Agents

Two editorial diagrams from the Substack — (A) how raw design artifacts become an agent-queryable wiki, and (B) the inside of a single Design Agent’s loop.

RAW SOURCES — immutable material INGEST → LLM WIKI — Karpathy, n.d. ↔ QUERY UX AGENTS — focused workers Product Spec / Features PRD / FEATURE LIST User Journey JOURNEY MAP Module List MODULE LIST Page List PAGE LIST User Flow USER FLOW LLM Wiki — LLM-maintained markdown index.md Catalog CATALOG log.md Chronological log CHRONOLOGICAL persona/ · decisions/ Curated pages CURATED PAGES cross-ref · backlinks · tags THE THING THAT DEGRADES IF UN-MAINTAINED Schema / CoT — writing rules · configuration Cross-ref conventions · file naming · lint rules · voice & tone Change the schema once → affect every page ↑ LINT · periodic audit ↑ Design Agent Draws UI · ships prototype Research Agent Interviews · ships insight Copy Agent UX microcopy · voice Red Team Agent Contradictions · adversarial Orchestrator = YOU · the curator after Bush (1945) · Karpathy (n.d.)
Ingest · Feed material New PRDs, journey maps, and module lists arrive — AI reads them, summarizes, and builds cross-references. Humans decide what’s worth keeping. Specs that lived only in documents become knowledge agents can query.
Query · Read + Write Agents pull context from the wiki (no more re-briefing every session); valuable answers drop back as new pages — chat history isn’t knowledge, the wiki is.
Lint · Audit Periodic automated checks for cross-page contradictions, orphan pages, and missing links. Rules live in the schema — change one rule and every page is affected.

HCI/UX Community Co-Creation

Co-Creating an AI / UX Book

6-month sprint · ongoing

Reading-Group Alumni Co-Creation · hosted on Patreon

  • Continuing book-creation initiative built with alumni of the UX For AI Reading Group — 22 chapters completed in 6 months, structured as a two-sided manuscript.
  • Side A — UX Practice Path (AI in products): a linear “AI Literacy Formation Journey” — chapters that walk a practitioner from foundations to applied product decisions.
  • Side B — Management Strategy Path (AI in your team): a goal-oriented “AI × Gen-Z Co-working Playbook” — strategy chapters for leaders managing AI-augmented Gen-Z teammates.
  • Membership-supported on Patreon with an active trial-reader feedback loop — drafts iterate against real-reader reactions, not in isolation.
22 chapters completed 6 months of co-creation 90+ trial readers
Manuscript Preview
AI/UX Doing book co-creation — preview 1
AI/UX Doing book co-creation — preview 2
AI/UX Doing book co-creation — preview 3
Join on Patreon

UX For AI Reading Group

6-week cohort

UserXper · Circle Community

  • Designed and delivered a 6-week applied reading group exploring UX patterns, governance frameworks, and design critiques for AI products — bridging academic HCI literature with practitioner workflows.
  • Hosted live discussions and reading-circle sessions on the Circle community platform, with weekly assignments and group critiques.
228 cumulative participants 6 weeks of live instruction Circle community platform
Teaching Material Preview
UX For AI Week 5 — Three Pillars of AI-Driven Design
UX For AI — slide overview grid 1
UX For AI — slide overview grid 2
View Course Watch on YouTube

Consulting Cases

McKinsey-lens UX consultancy AI breakthrough — slide 1
McKinsey-lens UX consultancy AI breakthrough — slide 2
McKinsey-lens UX consultancy AI breakthrough — slide 3
McKinsey-lens UX consultancy AI breakthrough — slide 4
1 / 4

A McKinsey-Lens View on the AI Breakthrough for UX Consultancies

Convert the tacit wisdom behind design work into AI-queryable data assets — and UX consultancies can build a stronger moat in the AI era than traditional large-format firms. The case maps where the proprietary judgment actually lives, and how to encode it.

Client UX Design Consultancy — Industry Advisor
GOV.UK to Taiwan UX transformation — cover
FRAME reference card — Framing, Remembering, Actionable, Meaningful, Evidence
How Might We #1 — translating data-driven design into government workflows
How Might We #2 — building a Taiwan-native design system, not a GOV.UK copy
1 / 4

From GOV.UK to Taiwan — A Government UX Transformation Framework

How UK’s GOV.UK / NHS design practice translates into an executable framework for Taiwan in the AI Basic Law era. Direct replication of British minimalism fails — the case rebuilds the design system around Taiwan’s actual context: aging users (80+), multilingual realities (Mandarin / Taiwanese / Hakka / Indigenous languages), and AI transparency requirements. Anchored by a FRAME reference card (Framing · Remembering · Actionable · Meaningful · Evidence) adapted from Chih-Yuan Experience Design (2025).

Client Public-Sector Digital Strategy — Government UX Advisor
Copilot to Agent to Multi-Agent — three stages of AI control transfer
Multi-Agent UX design patterns — role attribution, status, handoff, human gates
Multi-Agent experience evaluation — transparency dimension
Context window structure, context rot, and compact strategies
1 / 4

Agent Governance & Multi-Agent UX — Financial Services Executive Curriculum

Executive workshop curriculum on turning AI from chatbox pilots into governed, ROI-positive production systems in regulated financial services. The arc moves from Copilot → Agent → Multi-Agent control transfer (each stage with different trust and oversight assumptions), through four frontend patterns—role attribution, status display, transparent handoff, and human intervention points—so users can understand, control, and trust delegated work. A five-dimension experience scorecard (control, transparency, consistency, fault tolerance, efficiency) benchmarks global banking AI deployments; transparency is framed as explainability under compliance, not optional UX polish. The closing module treats the context window as operational “working memory”—context rot, compact/continue/rewind/clear/subagent strategies, and token budgeting as joint AI/UX governance levers.

Client Regional Banking Group — AI/UX Executive Workshop
Web4 agentic economy — market size and momentum forecast
Human oversight three-layer framework — In, On, and Above the loop
Breakout readiness — friction, aha moment, and viral sharing conditions
Visible-contribution affordance design — attribution, judgment explanation, reflection triggers
1 / 4

Web4 Agent Economy — Oversight, Breakout Readiness & Visible Contribution UX

Strategic UX advisory for a Web3 personal-agent platform shifting from passive on-chain identity to paid, delegated labor in a Web4 agentic economy. The analysis anchors market tailwinds (agentic AI toward ~$199B by 2034; enterprise agent embed rates accelerating 8× within 12 months), maps human supervision across In / On / Above-the-loop control, and isolates the decisive gap: Epistemic Access — users cannot see which profile dimensions drove an agent’s call, which breeds illusory control at the strategy layer. A breakout-readiness scorecard tests three simultaneous conditions—frictionless onboarding, a first real payout that validates intuition, and shareable proof of judgment quality over credential display. Recommended design moves favor visible-contribution affordances (payout attribution, judgment explanations, reflection prompts on atypical accepts) over deeper algorithmic disclosure, following Rismani et al.’s reflection-oriented principle.

Client Web3 Personal-Agent Platform — Strategic UX Advisor

Past Clients / Partners