Field Notes · Technical Strategy

How to choose a tech stack built to last a decade.

A founder's guide from the studio: how we frame the stack as a foundational design decision, not an engineering afterthought — and how that framing separates products that survive their category from those that don't.

The stack is a design decision

When founders ask us how to create a digital product, they usually expect a conversation about wireframes and brand. But every interface decision — how fast a page loads, how a form recovers from error, how a feature ships six months from now — is downstream of the stack. A tech choice made in week one dictates what your team can do in year five. That's why our technical strategy work sits alongside interface architecture, not behind it.

What "longevity" actually means

A product design agency that promises a decade-long product is really promising four properties of the underlying system:

  • Portability. The code can leave any vendor. If a hosting provider, database, or framework disappears, the product survives the migration.
  • Legibility. A new engineer can read the codebase and be productive in a week. Longevity is a function of how many humans can safely change the system.
  • Boring dependencies. The core of the stack is technology that will be around, unglamorously, in ten years. Novelty lives at the edges.
  • Reversible choices. Every decision is made assuming it will be wrong eventually. The cost of undoing it is small.

A framework we use with founders

Before we recommend a language, a database, or a hosting model, we walk founders through five questions. The answers usually make the stack self-selecting.

  1. 1. Who will maintain this in three years?

    If the answer is "someone we haven't hired yet," pick tools with a large, boring hiring market. TypeScript, PostgreSQL, and a standard React framework beat a niche stack, even a technically superior one, on this axis alone.

  2. 2. What is the smallest surface area that solves this?

    Every service, queue, and microservice is a future migration. Start with a single application, a single database, and a single deploy target. Split later, when a real constraint forces it.

  3. 3. Where does the data live, and who owns it?

    Data outlives code. A schema in a portable, open database is the most durable asset in the product. Treat it that way — even when the surrounding framework is trendy.

  4. 4. What is the exit cost of each vendor?

    For every managed service, write down what it would take to replace it. If the answer is "a rewrite," you have a lock-in problem you'll pay for later. If the answer is "a weekend of work," you can adopt it freely.

  5. 5. Is the framework's community still shipping in five years?

    Look at the last five years of the framework's release history, its funding, and who's employed to work on it. Longevity is a proxy metric for social durability, not technical merit.

What we reach for by default

For most products we build, the defaults look boring on purpose: TypeScript on the client and the server, PostgreSQL as the system of record, a single well-scoped React framework, a managed platform that we could leave in a weekend. We use PHP where its ecosystem is still the shortest path — content-heavy sites, mature back-office tools — and JavaScript everywhere else. The choice is never ideological; it's a bet on which stack we could still hand off to a new team in year seven.

The trap: optimizing for the launch

Most stacks look identical at launch. The differences show up in month eighteen, when a founder wants to add a second product line, hire a second engineer, or move off a hosting provider that raised prices 4×. A product design agency that only optimizes for the first release is quietly optimizing against the founder's next three years.

How we work with founders on this

Technical strategy is one of our disciplines, alongside interface architecture and full-stack execution. In practice, that means: we won't ship an interface without also owning what it runs on, and we won't recommend a stack we wouldn't still recommend in a decade. If you're planning a product and thinking about the next ten years, not the next ten weeks, that's the conversation we like having.

© 2026 — GENTLE OAK DESIGN AND DEVELOPMENT