You’re over-simplifying scaling. What is reliable and what is secure depends on what your software requires versus the costs. They are 2 different instruments, part of a large orchestra.
Apples to oranges. Traditional coding isn’t outdated, it’s foundational. Vibe coding isn’t a replacement, it’s an accelerator.
Cars eventually replaced horses because they were basically cheaper and faster.
There is a place for AI in software development but it’s just a tool and another stack. Like any tool the reaped benefits depend on the user’s ability to use it. Like any tech stack, it requires implementation towards continous delivery.
1 Like
If vibe coding becomes good enough, why would that somehow be disconnected from Bubble? Bubble stands to benefit proportionately, with the only difference being that the IDE is visual instead of text-based.
2 Likes
If you need to do that with AI code, and would need to do that with bubble app, what’s the benefit of AI coded app?
1 Like
Yellow is the by far best color
1 Like
My experience with vibe-oding tools vs Bubble.io
Bubble.io vs Vibe-coding
Disclaimer: I’m an ex-C++ with a past Assembler hobby, so I may be biased (verbal vs visual thinking)
Also, I’m into the startup/MVP building since 2011, so I may choose agility and user friendliness over hardcore strength.
I’m not a blogger, I’ve been asked to keep journaling my journey with this app, so pardon my style, it’s my 3rd post/article in my life.
2 Likes
I have built a service in Python/Django etc. stack + TS on frontend using Claude Code. Not vibecoding but using intentional coding, designing database logic, functions, flows etc. Setting up server, tunnels, security is not that hard with the avaialbility of the frameworks.
As I like bubble in general - for 2 years from my perspective - it is standing in a place.
No loops, proper flows.
No transactional locks.
No cheaper EU Hosting and security for some high risk elements (as you can see from some posts, logs are still in US)
Slow nested searches, no advanced search logic on the serverside.
Old and unreliable plugins (PDF generation for longer PDFs, a lot still require external services, mass of plugins are very shaky and unreliable and building a plugin is not a pleasant flow).
API Call config is inferior to what you can do with code - imagine how easy it is to setup a proper recalls, other paths in code vs what we have in bubble.
Illogical and unsequential workflows.
Bad native logs (a lot of actions are missing).
RG Pagination is dasterdly to build for some of the frontends.
No fuzzy search.
Limited syntax (ex if you want to sutract dynamic value from 1 you need to cache 1 in the backend worklow, thats really limiting).
Repeating groups limitations of different kind. No Pivot tables.
A lot of weak components (searchbox value can’t be fully manipulated, date selector cannot handle input etc.)
I even need a workaround for OAuth logins due to limitations of bubble not being able to handle long MS Tokens.
I know a lot of workarounds, but I shouldn’t have to workaround everything..
I think, if the intentional coding will go as it is, bubble will slowly die out. Not because what they are not doing today, but what they haven’t done long ago - and built a lot of false promises.
I loved the all rounder approach of bubble and the possibility to build big and fast. It’s still there, but diminishing with every day of no changes actually done.
8 Likes
