AI Product ManagerNavigate lines between probabilistic and deterministic.

I take AI that already works and ship it into real budgets, real compliance and R&D teams that apply AI-native workflows.

Distinct people (one PostHog person per browser/device) with at least one pageview, since this site started tracking. No 30-day window on this one.All-time visitors69
Average session length, counting only sessions of 10+ seconds — quick bounces are excluded so a one-glance drive-by doesn’t drag the average down.Avg. visit0:37

01Case studies

Works

Shapped and Shipped 3 AI products.

Beever Atlas

Enterprise AI memory layer ·

An enterprise agent memory system that turns scattered team chat and personal data sources into a wiki.

AI PMGTMPricing
Beever Atlas product UI

Xprop (大廈通)

PropTech / property management ·

A WhatsApp-native layer over Hong Kong building management — Vision and ASR do the reading; a person still owns the price and the compliance call.

PropTechDocument AIPricing
Xprop's WhatsApp-native building management interface

V-Note Suite

AI meeting intelligence / GovTech + SaaS · 2025 — present

Cantonese meeting AI, sold two ways: air-gapped V-Note for regulated buyers, SaaS V-Note Live for everyone.

AI PMASRGTM
V-Note Live's meetings list, including recordings of Votee's own products

02Aspects we are covering

From Innovation to Practical Reality

Navigating the line between precise logic and the pragmatic calls that drive real deals.

01

Deterministic

The solution must fit the real-world use case: Compliance, accounting, law, RBAC, medical. Even 0.001% uncertainty is not acceptable.

02

Gray zone

Where strategy, decisions, human experts, actual scenarios and expectation management smoothen out the product delivery.

03

Probabilistic

Maximize the AI capability — adopt the latest technical breakthroughs and community innovations.

03How I work

Two worlds, one toolbox

I use both manual and agentic tools, whichever deliver the results.

The traditional half

Classic tools are there for a reason.

  • ExcelAccessible for all AI-readiness teammates.

  • FigmaWhere I think through ideas and collab with teams.

  • NotionMy digital notebook, extra context source for AI Agents.

  • LinearPM platform for roadmaps, milestones, and issues.

  • JIRANo choice if we have to use. Enterprise standard.

  • MiroDiscovery workshops for design thinking, team sync.

  • FramerFast, shareable prototypes when a static page is enough.

  • WebflowMarketing sites shipped without spending a dev sprint.

  • Adobe Creative SuiteBrand, print, and the graphic-design.

  • Optimal WorkshopRan Treejack tree-testing for UX study.

  • HotjarHeatmaps and replays — where the UI actually loses people.

  • SlackTeam chat, and Slack came from a gaming company?!

  • Google AnalyticsFunnels and drop-offs to check a design decision.

  • TableauBI dashboard and funnels at a glance.

The AI-native half

The new LEGO bricks I keep learning.

  • Claude CodeMy day-to-day helpers.

  • CursorTread this as an IDE place for repos.

  • MCPConnect to different sources for AI Agents.

  • Beever AtlasEat-your-own-dog-food AI memory system.

  • PostHogProduct analytics for the things we shipped.

  • Storybook CLIDesign system integrated with AI workflows.

  • AWS CLILet CC help me deploy faster.

  • GWS CLILets Claude Code reach my Google Workspace.

  • LovableQuick and handy prototype option.

  • PlaywrightBehavioural driven testing for our products.

  • LangfuseLLM observability. See what the models actually did.

  • OllamaEdge-device LLM let me work on planes. And free

  • StitchSpins up design.md / UX-UI concepts before I touch Figma.

  • DeepwikiPoints me at OSS repos and actually explains them.

  • GLMCheap tokens for the simple stuff, saves the good model.

04Community

Recent events

  • Global Service Jam 2026Global Service Jam 2026HelperMar 2026
  • Lovable — She BuildsLovable — She BuildsFacilitatorMar 2026
  • Votee — Vibe Design Thinking WorkshopVotee — Vibe Design Thinking WorkshopMain HostApr 2026
  • AI Tinker — A2UI & AG-UI HackathonAI Tinker — A2UI & AG-UI HackathonParticipantJun 2026

05Skills

The full inventory

Most of my skills at a glance.

01

Product

  • Product discovery
  • Product delivery
  • Roadmapping & prioritisation
  • GTM & pricing
  • Product analytics
  • Data-driven development
02

AI knowledge

  • AI agents
  • AI memory
  • LLM
  • Agent skills
  • A2A
  • MCP
  • RAG
  • Model metering
  • Agent swarms
  • ASR / TTS
  • Ollama
  • LM Studio
  • AG-UI
  • A2UI
  • AEO
03

Vibe-coding

  • Claude Code
  • Cursor
  • GitHub Actions
  • Oh-my-claudecode
  • Playwright
  • Lovable
  • v0
  • context7
  • deepwiki
04

Tooling

  • Linear
  • Notion
  • JIRA
  • Confluence
  • PostHog
  • Tableau
  • Google Analytics
  • Miro
  • Figma
  • Storybook
  • Webflow
  • Hotjar
  • GitHub
  • Docker
  • Vercel
  • AWS CLI
  • GWS CLI
05

Coding

  • HTML
  • CSS
  • JavaScript
  • Python
  • React.js
  • React Native
  • Git
  • Mermaid
06

Design

  • Design systems
  • User research
  • Usability testing
  • Wireframing & prototyping
  • User flows & journeys
  • Interaction design
  • Responsive web design
  • WCAG accessibility
  • A/B testing
07

Languages

  • Cantonese (native)
  • English (fluent)
  • Mandarin (fluent)