How one developer built a $1.5M SaaS inside the prison system

Executive overview

Most founders chase the same markets. Jordan Rejaud found a customer nobody else was building for — incarcerated people — and built a profitable SaaS in month one.

Paracute Chat is an AI chat service for prisoners, delivered entirely via the prison internal email system. Families on the outside pay $15–20/month; people on the inside get access to AI for legal research, learning, and family communication. No app store. No UI. Just email.

The core insight: the best niche is the one everyone overlooks because it seems too hard.

Finding the idea

• A client was unexpectedly sent to prison mid-project; Jordan stayed in touch via paper letters
• The client described prison services as low quality and overpriced
• Jordan saw a gap: apply modern software to a closed, underserved market
• Prison's closed ecosystem made conventional validation impossible — so building the MVP was the validation

Building and launching

• Prototype built in one month; payment system added in month two
• Profitable from month two; 200 paying users in the first month
• Tech stack: TypeScript, React, Postgres, Redis, Auth0, Prisma, Docker
• No mobile app — the product lives entirely inside the prison email infrastructure

Growth: word of mouth in a closed system

• Word spread organically among incarcerated users — no ads, no content marketing
• Built an internal referral system: recruiter earns a free month of credits when a referred user signs up
• 30,000 total users — roughly 20% of the entire US federal prison population
• ~9 million messages sent; ~100,000 family connections enabled

Business model

• Users (prisoners) and customers (their families outside) are different people
• Families pay $15–20/month; yearly discount available
• 2025 revenue: $300K+; lifetime revenue: $1.5M

On validation

• Most founders skip validation because they don't want to invalidate their idea — it's emotional, not strategic
• Validation isn't a specific framework; the only requirement is willingness to let the idea die
• Validate fast, validate early, validate often — before falling in love with the product

On building and failure

• "Overnight success" doesn't exist — Jordan's decade of prior mistakes made this one work
• The fastest way to learn is to start a bad idea now, in a controlled way
• Tech stack choice is irrelevant; speed of execution matters more
• AI has become the effective tech stack: Jordan hasn't opened a code editor in 3–6 months