All Case Studies
Case Study

From a Lovable Prototype to a Production Mental Health Platform in 4 Weeks

A non technical founder had a landing page and a basic product built in Lovable. HouseofMVPs handled the product strategy and the build, and delivered a production platform with therapist matching, booking, chat and video, payments and admin, on TypeScript, AWS and Amazon Bedrock with a HIPAA aware architecture, in 4 weeks for $16,000 fixed. The client is not named at her request.

Client: Client name withheld at the founder's request

Timeline
4 weeks
Investment
$16,000 fixed
Key Result
Prototype rebuilt as a production platform on a HIPAA aware architecture

The Challenge

An AI app builder gets a founder to something that looks like a product quickly. The distance from there to a product that can responsibly handle mental health conversations is large, and most of it is invisible in a demo: how data is stored and who can reach it, what happens when something fails, and how the AI parts behave when a person in distress is on the other side. A non technical founder cannot easily judge that distance alone. She needed someone to make the product decisions with her, not only to write code to a specification she was not in a position to write.

Our Approach

We started with the product, not the code. Before anything was built we worked through what the first production version needed and what could wait, and wrote it down as a fixed scope with a fixed price. Then we rebuilt the product properly instead of extending the prototype: a Next.js web app and a Fastify API in strict TypeScript, PostgreSQL through Prisma, AWS for hosting, and Amazon Bedrock for the AI features so that model access sits inside the same cloud as the rest of the system. Because this is mental healthcare, the architecture was designed around health information from the start. Retrofitting that later is far more expensive than building it in.

What We Built

Product strategy for the first production version: what it must do, what it leaves out, and the order of work.
A production codebase in strict TypeScript, replacing the Lovable prototype: a Next.js web app, a Fastify API and a PostgreSQL database with 44 models.
Therapist matching, so a person is connected with a suitable therapist.
Booking for sessions.
Chat and video for sessions.
A knowledge base.
Payments.
An admin area for running the platform.
Hosting and infrastructure on AWS.
AI features served through Amazon Bedrock.
A HIPAA aware architecture, designed around the handling of health information from day one.

Delivery Timeline

Strategy and scope

Worked through the product with the founder and fixed the scope and the price in writing before the build started.

Architecture

Designed the system for health information from the start, on AWS with Amazon Bedrock for AI.

Build

Rebuilt the product as a production TypeScript codebase.

Handover

Delivered the production platform at the end of week 4.

Tech Stack

Next.js (App Router)
Web app
Tailwind CSS
Web app
TypeScript (strict)
Language
Fastify
API
Zod
API
PostgreSQL
Database
Prisma
Database
Amazon S3 + KMS
Storage
Amazon Bedrock
AI
Turborepo + pnpm
Monorepo
AWS
Cloud

Architecture

webApp

Next.js with the App Router, Tailwind CSS and TypeScript.

api

Fastify, with Zod validation on every input.

database

PostgreSQL through Prisma, with 44 models and every schema change managed as a migration.

storage

S3 compatible object storage: MinIO in local development, Amazon S3 with KMS encryption in production.

ai

LLM calls go through Amazon Bedrock, inside the same cloud as the rest of the system, with no tracking of the data sent in those calls.

codebase

A Turborepo monorepo with pnpm workspaces and strict TypeScript everywhere.

Security

design

HIPAA aware architecture: decisions about storage, access and data flow were made with health information in mind from the first day of the build.

auth

Sessions are self hosted in PostgreSQL and carried in httpOnly cookies, so session data stays in the platform's own database.

access

Role based access control that denies by default: a role can do only what it has been explicitly granted.

inputs

Every API input is validated with Zod before it reaches business logic.

storage

Files in production are stored in Amazon S3 and encrypted with KMS.

ai

LLM calls run through Amazon Bedrock with no tracking of the data sent.

scope

HIPAA aware describes how the system is designed. Compliance is an ongoing programme that belongs to the organisation operating the product, and a development studio cannot certify it.

Key Takeaways

An AI app builder is a fast way to show an idea. It is not the same as a product that can carry sensitive data.

A non technical founder needs a partner for the product decisions as well as the code.

In healthcare, design for health information on day one. Adding it later costs more than building it in.

A fixed scope and a fixed price let a founder plan around a known number.

Deliverables

A written scope agreed before the buildProduction TypeScript monorepo: Next.js web app, Fastify API, PostgreSQL schema with migrationsSelf hosted session auth with default deny role based access controlTherapist matching, booking, chat and video, knowledge base, payments and adminAWS deploymentAI integration through Amazon BedrockHIPAA aware architecture

FAQ

Frequently Asked Questions

Related Case Studies

Want similar results?

Book a free 15-min scope review. Your vision, engineered for production in 14 days. Fixed price.

Book Scope Review