Skip to content
Fieldwork

Case study · 2025–2026

Instinct: a dog enrichment app, designed and built in three months

Instinct is a mobile app that gives working-breed dogs the mental work they were bred for. We built it end to end — brand, content, mobile app, backend, marketing site, App Store launch — from concept to live product in three months. Here's how.

Four Instinct mobile app screens shown together — Prize Hunt, Selective Retrieve with the AI 'Why this works' panel, Scent Trail collection, and a Rowan profile with fulfilment score

The problem

The UK's favourite dogs are all working breeds. Most of them never get to do the job they were bred for.

Labradors, cocker spaniels, border collies — they were built to think, track, retrieve and solve problems for hours a day. According to Dogs Trust, 83% of UK dogs show at least one undesirable behaviour, from reactivity to separation anxiety. The charity took 40,000 calls in 2024 from owners struggling with their dogs, often working breeds whose needs weren't being met.

The standard advice — more walks, more fetch, more "exercise" — makes it worse. What these dogs need is mental work calibrated to their breed.


The approach

Lean. Opinionated. Shipped.

Most product builds start with a six-week discovery phase, three rounds of stakeholder workshops, and a design system designed for a team that doesn't exist yet. We did the opposite.

Week one: framework. Six core canine instincts (Scent, Chase, Herd, Retrieve, Dig, Solve), a weighted profile system that turns a breed into an instinct mix, a three-tier progression model (Foundation, Intermediate, Advanced).

Week two: brand and IA.

Week three: prototype.

Week four: testing with real users.

By week eight, we were in TestFlight. By the end of month three, we were on the App Store. No discovery deck, no innovation workshop, no two-day kickoff offsite. Just decisions, made fast, by people who had to live with them.


What we built

Five parts, one product.

01

The brand — identity, logo, voice

Wordmark, identity system, voice, and a complete visual language built from scratch. The single most important brand decision: position Instinct as a performance-and-fitness app for dogs, not a cutesy pet-tech product.

Reference points were Strava, Whoop and Nike Training Club, not BarkBox. Bold display type, dark mode, real photography of working dogs doing work. We deliberately avoided the pastel palette and cartoon mascots that dominate the pet-app category — working-breed owners take their dogs' development seriously, and the brand had to take them seriously back.

02

The activity library — 144 activities, three tiers, six instincts

Every activity follows the same template: a title, a one-line hook, equipment, three-step instructions, walk-equivalent rating, and a "Why this works" note grounded in breed instinct.

We wrote a 63-item equipment glossary. A labrador owner and a border collie owner get genuinely different daily drives, not reskinned versions of the same five exercises.

03

The mobile app — React Native, Expo, full state machine

React Native (Expo SDK 55), TypeScript strict, Zustand and React Query for state, Supabase backend, Claude API for personalisation. Onboarding (breed selection, behaviour signals, instinct profile generation), daily drive execution with timer state machine, activity completion with rating and notes, explore with search and filters, progress with weekly charts and streak tracking, multi-dog profile management.

Auth tokens in expo-secure-store, persisted state across launches, error boundaries, offline fallbacks, full test suite, EAS build pipeline, TestFlight distribution, App Store launch.

04

The marketing site — instinct-app.com

Six instinct landing pages, a breed quiz funnel with GDPR-compliant waitlist capture, App Store integration, custom typography, responsive across breakpoints. Built to convert mainstream dog owners, not specialist trainers.

The instinct-app.com marketing site shown on a laptop — hero reads 'Stop walking your dog. Start working their brain.' alongside a portrait of a brindle dog

05

The AI layer — Claude API, production-ready

A "Why this works" panel on every activity, personalised to the user's specific dog using a structured prompt that takes breed, instinct profile, behaviour signals and activity context. Cost-controlled with caching, with a graceful fallback if the API is unavailable.


The validation

Real users, one real test.

Before App Store submission, we ran a usability and validation test with UK dog owners. Every screen of the onboarding and daily drive flow tested with real users. Findings were synthesised, prioritised and refined before launch.

This is what The Build looks like.

If you're a founder with a product idea and the runway to execute it properly, we'd like to hear from you. Most engagements start with a Field Study.