Case study · AI product + full-stack systems

InnerLeaf

An AI-assisted reflection platform combining structured AI output, Supabase authentication, role-aware user/admin experiences, database migrations and Playwright acceptance testing.

Full-stack product · Public GitHub repository
Interactive visual stories

Concept visuals showing how reflection, structured guidance and product quality fit together. These are not actual user reflections or product screenshots.

Concept visuals · no client screenshots

Illustrative project visuals are available with JavaScript enabled. The case study below remains fully accessible.

Problem

Free-form reflection can be difficult to organise into useful patterns or next steps. A product in this space also needs privacy-aware user accounts, reliable structured output, clear admin boundaries and a safe fallback when AI output is incomplete.

My role

I worked across product UX, AI integration, data/auth architecture, role routing, admin tooling, validation and acceptance testing to turn the product into a coherent full-stack system.

What I worked on

Structured AI workflow

The reflection API uses Google GenAI, parses structured output and normalises categories, facts, interpretations, thought patterns, unmet needs and next steps before the UI uses them.

User / admin separation

Separate user and admin experiences cover reflection workflows, account management, metrics, feedback, users and system pages with route-aware access behaviour.

Quality & evolution

The repository includes Supabase migrations, role/ownership protections and Playwright acceptance tests for routing and access boundaries.

Proof of work

Verified stackNext.js 16.2.7, React 19.2.4, TypeScript, Tailwind CSS 4, Google GenAI, Supabase, Radix UI, Zod, Recharts, Motion and Playwright.
AI boundaryThe API does not expose raw model text directly; it parses and normalises structured reflection fields before the UI consumes them.
Role routingPlaywright tests verify admin routing, user routing, blocked admin access for normal users and logged-out redirects.
Database evolutionSupabase migrations cover reflection fields, feedback, ownership, profile roles, admin protection and account settings.
Product depthUser reflection paths, history, summaries, profile settings, feedback, admin users, metrics and system views exist in the repository.

Tools & capabilities

Next.jsReactTypeScriptGoogle GeminiSupabaseZodPlaywrightProduct UX

What this demonstrates

This project shows that I can work across a modern product stack: product thinking, AI integration, full-stack system boundaries, authentication, database change management and automated acceptance testing.

I describe InnerLeaf as an AI-assisted reflection product, not a medical or therapy product. The case study focuses on product architecture, system behaviour and technical evidence visible in the repository.
Next step

See the evidence, then see the fit.

View GitHubBack to portfolio