AI app builders are evolving fast; here’s what our team learned after testing four of the most talked-about platforms.
AI-powered app builders are having a moment. With promises like “Build an MVP in 30 seconds” and “Skip the dev team”, platforms like Lovable, Replit, Base44, and Famous are changing the way people think about launching digital products.
But are they actually usable? Do they produce quality? And are they tools for developers and designers or replacements for them?
Well, that’s what we’re here to answer. We asked members of our team - from developers and designers to innovation managers…
Here’s what we found.
A clean, AI-first experience where users describe what they want, and Lovable generates an entire app from it. Built on top of Supabase and Vercel, it promises developer-grade results without touching a line of code.
Replit has been around for years, but its latest updates combine a built-in IDE, code execution environment, and AI autocompletion powered by Ghostwriter. It’s less about no-code and more about enhancing dev workflows, especially for solo coders or learners.
Base44 positions itself as the AI startup cofounder. You describe your product vision, and it spins up a basic SaaS or landing page, complete with user auth, database, and styling.
Famous focuses on speed and beauty, marketing itself as a tool for generating landing pages and microsites with minimal effort. Think Webflow meets ChatGPT.
These platforms are built to democratise product creation but do they actually do the job for real-world teams?
AI app builders can be incredibly fast and powerful, but understanding how to use them effectively is key. As our Senior AI & Software Developer, Harry Townsend points out, these tools “will build literally what you tell them to.”
That’s where a skilled developer or designer adds value - guiding the AI toward solutions that truly meet the user’s needs. In Harry’s words: “A good developer will try to understand what you’re really trying to achieve and guide you in a better direction - even if it means adjusting your assumptions.”
When used thoughtfully, AI can accelerate development, help teams experiment quickly, and bring ideas to life in ways that were previously time-consuming.
As Harry explains, the real benefit comes from combining AI’s speed with human insight: it’s not about replacing people, it’s about making their work more effective.
“Modern development tools are simultaneously good and bad. The problem is that, when they're bad, an inexperienced developer (or non-developer) won't be able to recognise or understand that it’s bad."
These tools are still evolving, and as with any emerging technology, there have already been a few high-profile missteps.
In July 2025, Replit’s CEO apologised after its AI tool accidentally deleted an entire company’s production database. One user gave a vague instruction and the AI obliged, without any safeguards.
But as our CEO, Steve, put it: "Of course these are just people doing silly things! You can easily prevent the AI from having access to do this – and it could very easily run commands that cause all manner of hell. That's not to say you should not use it of course!"
It highlights a key divide: tools designed for developers vs tools for "vibe coders." Some platforms hide the code entirely and expect you to prompt your way through. Others, like OpenAI’s Codex, integrate directly with GitHub and can raise pull requests, working more like a collaborative junior dev.
Steve adds "I haven’t seen anyone launch an open-source version of Lovable, Bolt.new, or Replit yet."
That’s one reason we built our own internal UX/UI generator. Not for full app builds, but to generate reliable HTML components in the right context. Easier to test, quicker to build on, and a smarter way to accelerate development.
The growing consensus across our team: technology, especially AI, is a tool, not the solution.
Something our Innovation Strategy Manager, Tom, has been talking to clients about is reframing how we look at technology, especially AI, and shifting the perspective. He mentions how "framing AI as a creative collaborator rather than a full solution seems to resonate. Humans and organisations who collaborate with AI now will replace those who just use it."
With Mark, our CIO adding "It’s a co-pilot, not a human replacement. People / Process / Technology: tech is the tool, people are the solution."
AI should be seen as an enabler of faster, more creative workflows. But human oversight, expertise, and intent still steer the ship.
Harry’s earlier point touches on a nuance most people miss.
"AI can be a powerful accelerator but only when you’ve built the mental models to use it well. The ability to read and understand code quickly is more valuable than simply writing it. But you only get there by going through the painful trenches: writing code, breaking things, fixing them, and learning how systems behave. That’s how you build the instinct to emulate the computer in your head.” adds Steve.
In other words: the more you know, the more useful these tools become.
These tools are fast, fun, and (in some cases) surprisingly functional. For prototyping, validating ideas, or building internal tools, they offer a powerful shortcut.
But they’re not a replacement for thoughtful product design, strong architecture, or commercial strategy. They’re a launchpad, not the entire flight path.
Great for:
Risky for:
At New Icon, we help teams turn ideas into intelligent, scalable products using the best of AI, human creativity, and solid software engineering. Talk to us about building your MVP.
Subscribe to get our best content. No spam, ever. Unsubscribe at any time.
Send us a message for more information about how we can help you