← All work

Own product · SaaS · 2025 – Now

Wise — AI Operations Platform for SMEs

From 0 to 1: product strategy, UX and full-stack build of an AI platform for Hong Kong service businesses

Role
Product designer & full-stack developer
Timeline
2025 – present (ongoing)
Team
Solo — strategy, UX/UI, frontend, backend, AI, DevOps
Type
Own product · SaaS
Wise — AI Operations Platform for SMEs

01The problem

Hong Kong SMEs in service industries — beauty salons, TCM clinics, tutoring centres, pet grooming — run bookings, payments, customer chat and loyalty across disconnected tools. Most still rely on phone calls and WhatsApp, leading to missed enquiries after hours, double bookings and no data visibility.

Existing SaaS is either too generic (not localised for FPS/PayMe and HK business practices), too expensive, or too technical for owners with no tech background. The bar was a platform powerful enough to replace five tools, simple enough to set up in five minutes.

02What I did

Product strategy

  • Researched nine HK service verticals and mapped shared workflows versus industry-specific ones
  • Defined a modular architecture: shared core (AI chat, bookings, payments, CRM) with pluggable modules per vertical
  • Designed a four-tier pricing model from solopreneurs to multi-location chains

AI engineering

  • Built a RAG conversational engine that ingests PDFs, Word docs and FAQs as merchant knowledge bases
  • Multi-model LLM support with a customisable system-prompt framework per merchant
  • WhatsApp Business API integration so merchants answer customers where they already talk

UX and interface

  • Three-step onboarding — upload knowledge, copy widget, go live — validated under five minutes with non-technical users
  • Admin dashboard with real-time conversation analytics, booking calendar and FPS reconciliation
  • Industry-specific UI patterns: treatment records for clinics, pet profiles for grooming, attendance for tutoring

Full-stack build

  • React front end, Node.js API, PostgreSQL, Redis and WebSocket-based real-time chat with AI-to-human handoff
  • Multi-tenant architecture with white-label support and local payment integration (FPS, PayMe)
  • Membership engine with points, referrals, tiers and no-show blacklisting

03The outcome

Live and serving active merchants across multiple HK service verticals
AI handles 80%+ of routine customer enquiries autonomously
Under-five-minute onboarding validated with non-technical users
Bookings, payments and loyalty run on one data layer — no more silos

05Tools & stack

React TypeScript Node.js PostgreSQL Redis RAG / LLM WhatsApp Business API Docker Cloudflare

Let's build
something real.

Open to full-time roles · Hong Kong — or a conversation about hard product problems.

bendyyip@gmail.com