What the Flowuity build stack looks like, and why we made each call.

Next.js, Supabase, Sanity, Stripe + PayFast, HubSpot, Resend, Twilio, Anthropic. The reasoning behind a default stack — and the conditions under which we deviate.

Choosing a stack is a commercial decision before it is a technical one. The stack that ships the product, that the team can hire for, that the client can run after the engagement ends — that is the right stack. The stack that satisfies a particular engineering aesthetic is the wrong one.

Our default is Next.js with React and TypeScript on the front end. The reasoning is simple: it is the highest-quality frontend stack with the deepest hiring market, and it lets us ship server-rendered, fast, accessible products without a custom build pipeline. We use the App Router and Server Components by default.

Supabase for the data layer. Postgres, with row-level security, real-time, and authentication built in. The reasoning is that we get the long tail of Postgres — every problem you will encounter has been solved on Stack Overflow — without paying the operational tax of running it ourselves.

Sanity for structured content. We need a content surface that non-engineers can edit safely, that has versioning, that can be queried alongside our application data, and that does not lock the client in. Sanity is the cleanest answer.

Stripe and PayFast for payments. Stripe for cross-border, subscription, and the platforms that need them; PayFast for ZAR-denominated SA-only flows where it is materially cheaper.

HubSpot for CRM, with our Integration practice owning the data contract. Resend for transactional email. Twilio for SMS and WhatsApp Business. Anthropic Claude for AI features that need to reason.

We deviate when there is a real reason. A regulated client whose security posture forbids Vercel. A product whose latency budget rules out a serverless function. A team that already runs on a different stack and would rather we joined them. Deviation is fine. Deviation for novelty is not.