I’ve been building with WeWeb + Xano for over three years now, and it’s been a genuinely great ride. Watching both tools evolve and my own skills grow with them has been really rewarding.
But I’m hitting a real inflection point, and I’d love to hear if others are in the same boat.
The shift I’m feeling
With AI-assisted development accelerating fast, my workflow has changed significantly. On the backend side, Xano has made great strides. I can now drive a lot of my backend development directly through Claude Code, which is a game changer for speed and context awareness.
On the frontend side though, WeWeb is becoming a bottleneck in that workflow.
Here’s my current painful loop:
I design HTML mockups with Claude/Claude Code
I try to pass them to WeWeb Copilot → it fails more often than it works
I end up rebuilding the frontend manually in the editor, which defeats the purpose
This is pushing me toward an uncomfortable trade-off: rebuilding my full frontend in code (via Claude Code), which means trading the visual control and accessibility I love about low-code for raw speed and full AI context awareness.
The question I can’t answer yet
Is WeWeb moving toward giving external LLMs meaningful control over the editor? Something like what Webflow is doing with its MCP integration. It’s imperfect, sure, but the ability to fluidly switch between AI-driven dev and the visual builder is exactly the kind of hybrid workflow I’m looking for.
I’m not looking to bash the product. I genuinely want to stay. But I need to understand the roadmap well enough to make an informed decision about whether to continue investing in WeWeb or pivot my stack.
Would love to hear:
Is anyone else navigating this tension?
Any WeWeb team visibility on MCP or deeper LLM integration plans?
Hi, yes I feel it too, but for the moment I’m simply asking Weweb Ai to build my web apps pages by pages, based on a Claude description, and sometimes a pic (even if enclosing a pic makes Weweb AI struggling or shutdown) and it’s ok. And then I fine tune manually.
But I agree that it could be very smart to open to Claude MCP, or at least to improve Weweb AI
To conclude, I think you better stay on Weweb because you’ve mastered it. Don’t fall in the shiny object trap. You can actually deliver. Why change ?
I think the team is aware of this topic and I chose to trust them on it
I’m really hoping WeWeb AI will be able to help with visual building again soon. The previous version could do this, but for the past month we’ve been told “it’ll be back by the end of the week” — unfortunately that still hasn’t happened.
My project has stalled a bit because of this, since my workflow also relies on all three tools working together: Claude handles design and prompts, WeWeb builds the visual elements (section by section) and overall frontend, and Xano takes care of the backend.
I’m not asking for perfection — I understand things break. But it’s been over a month now, and the AI-chat is prominently featured on weweb’s landing page while it’s essentially unusable. That’s a bit of a tough sell. At minimum, I’d love to see reliable element generation back. Logic and workflows I’m happy to build myself — that’s never been the issue.
Hi @SOT I am happy you’re asking and the answer is yes, giving external LLMs meaningful control over the editor through MCP/CLI is firmly on the roadmap.
The hybrid workflow you describe is used by many of our users and we don’t want to be the bottleneck in that loop.
The reason we haven’t shipped it yet is that our AI and the CLI/MCP would use the exact same toolbox to build apps in WeWeb. So we need to make our AI reliable first otherwise your external LLM would inherit the same rough edges our AI is hitting right now. Fixing them once benefits both.
Good news is that with the new version now in production since 3 weeks, our pace of improvement is dramatically faster. We’re targeting a reliable AI by end of June, with CLI/MCP following right after.
Hi @Michael, thanks for your feedback! Have you tried AI since last Thursday? My latest tests on UI output were genuinely impressive. Can you try giving your prompt directly to the interface agent and let me know how it goes? Front-end building without logic has been really great on my side.
Super happy to hear that @Raphael !! I will wait a few months to see how it goes. Pricing wise since it directly compete with the current token based pricing, do you plan on making the MCP/CLI feature limited to certain plans ?
Pricing-wise we don’t really know yet tbh, but it will be accessible. I don’t think we’ll gate that behind a scale plan or smth if you know what I mean.
