Back/ News /

Can you build a bank in Lovable?

Av Daniel Berg

Wizardworks is responsible for backend, infrastructure and security at Zinova, the new bank building its frontend with the AI tool Lovable.

Today, Breakit and Dagens industri report that serial entrepreneur Tommy Jacobson is launching the bank Zinova — with the entire frontend built in the AI tool Lovable. Wizardworks builds the rest: backend, infrastructure and security. Everything required to operate in a regulated environment.

A new delivery chain takes shape

What's interesting isn't who builds what — it's how the new delivery chain works. Domain experts can use AI to rapidly create software that meets requirements. In concert with partners like Wizardworks — where we, together with our AI agents, build everything else: architecture, APIs, integrations, infrastructure and security. With agentic AI, we deliver faster and with higher quality, and more fits in the same budget.

Lovable is not a prototype

It's a requirements tool. Product owners and designers express exactly how the service should look and work — in actual code. The result is working views and flows that can be tested and iterated before going further down the rabbit hole. Lovable is just one of many alternatives. Figma Make, Bolt, Replit — a whole category of tools that let people closer to the business define what should be built, in a format developers can work with directly.

Production-grade in a regulated environment

In a regulated environment, looking good isn't enough. Code must be auditable, testable and production-ready. Architecture must meet regulatory requirements. Infrastructure must be reproducible and version-controlled. Security must be built in from day one. This is where the agentic delivery chain takes shape: design, specification, implementation, validation and deployment. Every step can be accelerated by AI — but every step also requires human judgment and accountability.

Three takeaways

Design tools like Lovable transform requirements gathering — when the business side can show exactly what they mean, misunderstandings decrease. It's not prototyping, it's specification. Production-grade still requires engineering — AI-generated code is a starting point, not a finished delivery. And the new chain is agents and humans together: design agent, code agent, test agent — but with human judgment at every critical step.

In the media

Building something that needs to be production-grade? Contact us and we will help you take ideas from design to production-ready delivery.

Daniel Berg

Written by

Daniel Berg