AI-assisted SaaS · multi-tenant

PupManager

Building an AI-assisted SaaS platform for dog trainers and behaviourists.

PupManager product screens

My role

Product design & build

Users

Dog trainers & their clients

Stack

Claude · Vue.js · Supabase

Type

Multi-tenant SaaS

Project overview

PupManager is a multi-tenant SaaS platform for dog trainers and behaviourists — handling scheduling, class registrations, client management and financial workflows. It was designed and built with Claude, Vue.js and Supabase using an AI-assisted product and development process.

The problem

Dog training businesses juggle bookings, recurring classes, client records, payments and calendars across spreadsheets, paper and disconnected tools. There was no affordable, purpose-built platform that understood how a training business actually runs day to day.

My role

I designed and built the product end to end — product strategy, UX and UI in Figma, the multi-tenant data model, and the implementation itself in Vue.js and Supabase, with Claude as an AI pair throughout. I also drove the website and go-to-market thinking.

Users and context

Dog trainers, behaviourists and their clients (dog owners). Trainers need to run classes and 1:1 sessions, take registrations and payments, and keep client and dog records. Owners need to find, book and pay for the right sessions.

Constraints

A solo build; multi-tenant from day one; real financial workflows; third-party integration (Google Calendar); and it had to be genuinely usable by non-technical small-business owners — on limited time and budget, which is exactly where AI-assisted workflows earned their place.

Process

I started from real customer pain at a dog training business, mapped the core workflows (scheduling, registrations, client management, payments), used Claude to explore product structure and edge cases, designed flows and screens in Figma, then built iteratively on Vue.js and Supabase — validating against real scenarios as I went.

Key design decisions

  • A drag-and-drop scheduler as the heart of the product, matching how trainers think about their week.
  • Multi-tenant architecture so each business has isolated data and its own branded experience.
  • Registrations and payments modelled around classes and sessions, not generic "products".
  • Google Calendar integration so trainers keep a single source of truth.
  • A deliberately simple admin surface for non-technical owners.

Outcome

A working, multi-tenant SaaS product designed and shipped by one person using AI-assisted product and development workflows — proof that senior product thinking plus AI can take an idea to buildable, real software fast.

What I'd improve next

Deeper reporting for business owners, automated reminders and retention flows, and a more guided onboarding that gets a new training business live in minutes.

How AI changed the process

  • Used Claude to explore product structure and edge cases early.
  • Used AI to draft flows, logic and technical approaches.
  • Used AI to speed up implementation thinking and unblock build decisions.
  • Used AI to move from problem to working product faster.
  • Still applied UX judgement and customer understanding throughout — AI accelerated the work, it didn't replace the thinking.

Want this kind of product thinking on your team?

See more workGet in touch