Pete Foster
Technology & Strategy Director
We’ve all been there. You see a demo of a new AI tool on YouTube or LinkedIn and, suddenly, the way you’ve been working for years feels very slow. It might even feel redundant.
We’re navigating a new era of prototyping, development and production. Between no-code platforms like Lovable, agentic sandboxes like Replit and AI-first integrated development environments (IDEs) like Cursor, the barrier between an idea and a working prototype hasn't been lowered. It’s been reduced to rubble.
But there’s a trap. If you try to use every tool at once, or use the wrong tool for the task, you won’t end up with a faster or better product; you’ll have messy, unmaintainable code.
Experimentation to expertise
Making sense of this landscape requires more than just reading documentation; we need to get our hands dirty. The only way to truly understand these tools is to work with all of them, testing their limits to see where they shine and where they stumble.
By combining constant experimentation with technical expertise, we can move beyond a wall of competing platforms and favourites, to work with specific solutions for specific outcomes.
The right tool for the job
We tend to categorise the current AI prototyping and building landscape into three groups. There’s grey space between them, but we like it like that.
-
The idea (e.g. Figma Make, Lovable):
These are the high-speed tools. If we need to show a stakeholder a functional UI design or a clickable concept by the end of a workshop, this is where we spend our time. It’s about look, feel, experience and immediate feedback. -
The sandbox (e.g. Replit):
AI coding agents help us to develop functional logic and test how data moves through a system without the overhead of a complex local environment. It’s the perfect middle ground for proving a concept and starting to think about architecture. -
The workbench (e.g. Cursor, Claude Code):
This is where we build. AI-driven IDEs don't replace software engineers or developers; they give them extra power. This is how we build scalable, production-ready software and websites that are integrated into secure, maintainable infrastructure.
When to switch
The real skill isn’t just in using the tools. It’s in knowing when to put one down and pick up another. A prototype built in a no-code tool is an immediate, effective communication device, but it’s rarely the final destination.
Speed is vanity if you’re heading in the wrong direction. We use these tools to fail fast, learn quickly and then, critically, to build things that last.
Want to play?
AI and automation are moving incredibly fast. It’s a truly exciting time to be in tech, but it’s also daunting and the unknowns are many. If you’re curious about how these tools could speed up your next project or you want to learn more, come and see us. We’ll be happy to chat, show you the ropes, or even start building an idea together.