The Solo Founder Tech Stack We Use for Every Product (2026 Edition)
The exact stack behind every FGA Labs product in 2026 — and why choosing it once, for everything, matters more than any individual tool.
By FGA Labs
People ask what stack we use expecting a list of tools. The honest answer is that the specific tools matter less than the meta-decision: every product we ship uses the same stack, chosen once, revisited rarely. When you run many products with a small team, the stack is not a technical choice. It is an operational one.
The actual stack
- Next.js with TypeScript for every web product — one framework, one mental model, everywhere.
- Postgres for anything that matters. Boring, proven, and the operational knowledge transfers between products.
- Edge-first hosting with static generation wherever possible — fast by default, cheap at low traffic, no servers to babysit.
- Stripe for payments, one shared integration pattern, lifted wholesale into each new product.
- AI agents in the development loop for scaffolding, migrations, tests, and refactors.
Why sameness beats optimality
For any single product, our stack is probably not the theoretical best choice. There is always a framework better suited to this app, a database tuned for that workload. But we are not optimizing a product — we are optimizing a portfolio. Sameness means a bug fixed once is fixed everywhere. It means any product can be picked up cold in minutes, because it is shaped like all the others. The compounding returns of uniformity beat the marginal returns of per-product optimization, and it is not close.
What we deliberately do not use
No microservices — a portfolio of monoliths is already distributed. No Kubernetes — we have nothing that needs it and no one whose job is tending it. No new database technologies, however compelling the benchmarks. Every technology you adopt is a technology you operate at 2am, and our entire model depends on 2am staying quiet.
Choose your stack like you will still be maintaining it in ten years with the team you have today — because you will be.
The stack also turns out to be the reason AI agents work so well for us. Uniform conventions across every repo mean an agent that learns one product can work on all of them. Boring choices, it turns out, are also the most automatable ones.