Get a free demo
Back to home

Case study

Subscription web app for an international client

From brief to production: an app with two types of users, subscriptions, real-time communication and content moderation, installed by people straight to their phones.

users in the first week of operation
1,000+
countries
70+
requests processed in 24 hours (API and cloud)
110,000+
API call success rate
99.8%

The brief

The client needed an app that connects two types of users who react to each other in real time. It had to work on mobile like a native app, earn through subscriptions and give the client tools to manage users and moderate content.

What we delivered

  • Real-time layer

    State changes reach everyone watching instantly over WebSocket channels (broadcast). User online status (presence), live chats and a live event feed on public pages.

  • Progressive web app (PWA)

    Service worker with offline cache, installable to the phone, push notifications via Web Push.

  • Subscriptions and payments

    Stripe Checkout, monthly and yearly plans, events processed through verified webhooks. Card data never touches our servers.

  • Hybrid rendering

    Public pages are prerendered or server-rendered with edge CDN caching; the private area runs as an SPA outside search indexes.

  • Admin panel

    Three permission levels, user management, report moderation, an audit log and a revenue overview.

  • Security

    Data access is enforced by the database itself (row-level security); writes go only through the verified server API. Rate limiting lives in the database, so it works in a serverless architecture too.

How we built it

We brought AI agents into every phase of development. Decisions stayed with a human.

  1. Specification

    We first described every part of the system in a structured specification. It is the context the AI agents work from.

  2. Implementation

    Agents write code from the specification and the task brief, within a clearly defined architecture.

  3. Review

    A language model automatically reviews every change in the CI/CD pipeline. A separate AI-assisted audit checked database and API security, and regression tests make sure the issues found stay fixed.

  4. Approval

    Only what passes the developer’s human review goes to production.

Technology

  • Nuxt
  • TypeScript
  • Cloud (PostgreSQL, real-time, auth, storage)
  • Serverless functions
  • Edge CDN
  • Stripe
  • Web Push
  • AI agents

Have your own idea for an app, or a process you want to simplify?

We'll show you how we would build it.

Get in touch