Your cloud bill grows faster than your product.
A service with 100 users a day runs on a cloud server for €30 a month. Add a database for €10, an analytics subscription for €10 and often a CMS on top. Each item seems small, but together you pay many times what the application really needs every month.
Many small projects are built with the toolbox of large platforms: a framework like React, Angular or Vue, a separate database, a CMS, an analytics service. Every building block costs money, needs updates and widens the attack surface. Yet the basics are so capable today that many of these layers are unnecessary. Go, Rust, JavaScript and CSS have been stable for years, and with AI agents you can develop quickly without a large framework.
A CMS is usually unnecessary for simple projects, too. You change content via chat, for example from Microsoft Teams, Claude or ChatGPT: an AI agent makes the change, a pipeline with automated tests checks it, and only then does it go live. That is faster than any CMS and just as safe.
What waiting costs you every month
Complex solutions built from frameworks, cloud services and dependencies don't just cost money every month. They tie your product to other people's decisions. When a framework or service reaches its end of life, you have to rebuild under time pressure. When a provider may no longer be used for legal or political reasons, your application suddenly loses its foundation. Every additional building block is another such risk.
AI agents make this worse. Without guidance they almost always suggest the same defaults, such as Vercel or AWS for hosting and a managed database like Supabase. For small projects that is often far more expensive without being better, and every further request adds another service.
Does this sound familiar?
A server, a database and several subscriptions run in parallel for an application with few users.
Every text change needs a CMS or a developer.
The frontend depends on a large framework whose updates regularly cause work.
How I approach it
Take stock
I record which servers, databases, services and subscriptions your application uses, what they cost and how much they are actually used. AI helps me analyze invoices and configurations quickly.
Simplify
I reduce the application to what it needs: a lean server in Go or Rust, an embedded database instead of a separate service, privacy-friendly statistics instead of an analytics subscription and a frontend of JavaScript and CSS without a large framework. AI agents do a large part of the rebuilding, architecture and review stay with me.
Edit content via chat
Instead of a CMS, you change texts and images via chat from Teams, Claude or ChatGPT. Every change runs through a pipeline with automated tests and only goes live once everything has been checked.
Frequently asked questions
Will our application get slower?
Usually it gets faster. Fewer layers mean less loading time and less that can fail. Every change is measured before and after.
So we don't need a framework like React anymore?
For many projects, no. Browsers can do much of what used to require a framework, and Go, Rust, JavaScript and CSS have hardly changed in years. That means fewer updates and fewer dependencies. For large applications with many teams a framework can still make sense, we clarify that when taking stock.
How do we change content without a CMS?
Via chat: you write in Teams, Claude or ChatGPT what should change. An AI agent makes the change, a pipeline with automated tests checks it, and you approve it. Every change stays traceable and can be undone at any time.
Is it safe when an AI changes content?
Yes, because the AI doesn't publish anything directly. Every change runs through the same tests as any other code and only goes live afterwards. Faulty changes get stuck in the pipeline and never reach your users.
Which providers do you work with?
With the common cloud providers such as AWS, Azure and Google Cloud, as well as Kubernetes and traditional servers. For small applications a single inexpensive server is often enough, ideally with a European provider.
Can't AI simplify our application by itself?
AI rebuilds quickly, but without guidance it reaches for the familiar: Vercel or AWS, a managed database like Supabase, the large framework. That is convenient, but for small projects often expensive without being better. What your application really needs and what can go, I decide together with you. AI agents do the implementation under my guidance, backed by tests.
Let's talk about your situation
In a free initial call we take a look at your problem together. You get an honest assessment of whether and how I can help.