I started a competition with my friend today to see who could build the best app. Not necessarily the fastest build but the best build. The challenge?
I will build the app on bubble. He will build the same app with claude.
He’s super confident that claude can build it but I don’t think it can. I’ve seen so many terrible apps built with claude and bubble gives me everything I need to build the app myself so why even bother with claude for it?
This will be my first time using bubble and it will be his first time using claude. Neither of us are coders but we both have started a business together to build this app to solve a problem at our work.
The challenge will last 1 week and then we will see who the winner is.
Anyway, I wanted to share so you guys can give me tips to build it correctly. And also to see if any of you have run a similar experiment before?
The app is a recipe costing tool for the restaurant we work at - the menu changes daily and currently we have to manually figure out how much it costs in ingredients and time for each menu item. We tried using AI for it but we need to be exact with the ingredient costs not just a rough estimate.
A few key features will be recipe costing, ingredients lists, recipe library, recipe ratings and a history of last time we cooked a recipe (we try not to reuse the same recipes too often!)
I read that bubble uses claude for the ai builder so I guess this isn’t really a fair comparison! I’m still waiting for the AI builder to do its thing. Does it usually take a long time?
On the “Bubble uses Claude so it’s not fair” point, the AI builder inside Bubble isn’t the same thing as asking Claude to generate and run the whole app for you. It’s scaffolding inside a structured system that isn’t wired to anything.. Very different from freeform AI trying to architect everything.
For what it’s worth, I actually built a quick version of a recipe costing tool today fully in Claude while stuck on a conference call. No Bubble involved.
It’s already handling:
Ingredient-level unit costing (deterministic math, not estimates)
Recipe → ingredient joins
Daily menu view
Station breakdown
Cook history + ratings
The interesting part of your challenge won’t be who gets something working first. It’ll be whose system stays clean once you have 200+ recipes and real pricing changes.
If I were you, I’d focus on the data model first. For costing tools, structure > UI every time.
Curious to see how it plays out. Keep posting updates.
Keep us posted, I think you both will get pretty far, but a complete app might be difficult with just a week. Would coaching help void your competition?
Great challenge! I am actually also in the process of trialing Claude (with a few supporting tools) to see how it compares to Bubble on smaller projects. Will stay tuned to see how you guys progress on this. Best of luck.
