Based on the question asked by @boston85719 in this other thread, I decided to create this thread with my thoughts to maybe start a constructive discussion about it.
I think there has been a considerable drop in opportunities that previously required Bubble as a tool. I noticed it gradually over time, although I couldn’t quantify it precisely.
Back then, I used to receive several Bubble job offers through LinkedIn PMs and often shared them with other developers in my network and whatsapp groups, since I was already working full-time with Bubble.
Sometimes I actively searched for Bubble opportunities just to share them with others. There used to be many. Today, genuinely good opportunities feel much rarer. Freelancer platforms also seem heavily affected.
From my perspective, there are four main groups of factors involved:
- AI related hype and expectations
- Bubble related decisions and perception
- Comparisons between Bubble and AI-first tools
- Market saturation and low quality labor
GROUP 1 — AI Hype
Every day, the internet pushes the idea that anyone can now build complete apps instantly using AI, regardless of technical knowledge.
Many of us already know this is heavily exaggerated. Serious limitations still exist, especially due to the lack of technical knowledge required to properly handle things like architecture, scalability, security, and maintainability. But the marketing narrative is extremely powerful.
I believe many founders decided to either:
- try building things themselves with AI, or
- hire people promising “full apps in days”
And in many cases, Bubble stopped even being considered. Which brings us to the next group of factors.
GROUP 2 — Bubble’s Direction and Priorities
When AI exploded, almost every platform suddenly decided it needed to be associated with AI somehow. Some had genuinely relevant use cases. Others seemed mostly interested in “surfing the AI wave”. Bubble was one of those platforms that naturally moved in that direction as well.
And to be fair, there are genuinely valid reasons for integrating AI into Bubble. Buildprint itself demonstrates that clearly. But personally, I think the execution and prioritization may not have been ideal.
I’m not part of the Bubble team, so I obviously don’t have the full context internally. But from the outside, it sometimes feels like AI initiatives absorbed too much focus while many long requested core improvements and QoL (Quality of Life) suggestions remained untouched for years.
At the same time, I also believe a lot of what @georgecollier built with Buildprint (BP) could have influenced Bubble much earlier and more deeply in certain areas. In my opinion, someone with that kind of practical understanding of AI-assisted Bubble development is exactly the type of person I would want closely involved in shaping Bubble’s AI direction.
Maybe conversations like that already happened internally. Maybe there is already collaboration happening behind the scenes that we simply don’t see. I genuinely don’t know. But, for example, it’s hard not to notice that some things introduced into Bubble more recently seem at least partially influenced, directly or indirectly, by ideas that Buildprint 1.0 had already been exploring for quite some time. Like some things added to the security dashboard. And that’s great!
And to be clear, this doesn’t invalidate the work that the Bubble team has already done with Bubble’s native AI efforts and other things. Not at all.
I believe that when an industry shifts this quickly, bringing together some of the strongest and most forward thinking people in that specific space becomes extremely valuable, especially while the momentum, excitement, and willingness to innovate are still there.
But again, @georgecollier also has his own projects, his own agency, and we don’t really know the actual level of involvement or cooperation happening behind the scenes around all of this. So maybe I’m just talking nonsense here.
That said, and as I’ve already mentioned in other discussions, the native AI experience is still inherently limited by Bubble’s current architecture, editor structure, native components, workflow logic, and other underlying limitations.
For example:
- workflow branching with proper if/else structures
- true loop systems
- backend workflow return data
These aren’t things plugins can fully solve elegantly, because the platform itself doesn’t natively support them properly. So naturally, Bubble AI also inherits those same limitations.
To be clear, and I think we can all agree on this, the Bubble team has shipped valuable improvements:
- API tab and API improvements
- Workflow tab improvements
- Security dashboard
- Privacy rules updates (preparing to future improvements)
- New property editor
- Mobile editor efforts
- Mobile plugin editor (in beta)
- and other things
These things absolutely matter, but I still believe Bubble would benefit greatly from having a small team dedicated exclusively to fast iteration on smaller QoL improvements and other pain points flagged by the community over the years.
You know, small but consistent improvements every week can significantly improve community morale and perception of platform growth and attention to community. As example, we had two Bubble Boost periods (if I’m not mistaken) in July and December of 2024, and those were very well received by the community. Imagine that kind of thing happening consistently every week, month, or year.
Because despite those efforts, frustration has also gradually grown among developers over time:
- delays
- bugs
- unfinished beta features/components (table component as example)
- releases perceived as lacking QA
- long-standing requests remaining unresolved
And naturally, those frustrations started spreading publicly across forums, social media, communities, and developer groups.
That inevitably affects how both developers and founders perceive the platform. And naturally leads into the next group of factors.
GROUP 3 — Bubble vs AI Tools
Years ago, Bubble was mostly compared to platforms like Webflow, WordPress, Maker, Framer, or FlutterFlow. Today, Bubble is increasingly being compared against a completely different category of tools.
On one side, we have AI powered builders/wrappers such as:
- Lovable
- Bolt
- Replit
- and other AI-first builders/wrappers
And on the other side, Bubble is now also being compared directly against coding-oriented AI models like:
- Claude Code
- Codex
- Cursor
- and agentic coding workflows in general
And those comparisons are often unfair or incomplete because most people ignore factors like:
- technical knowledge requirements
- security risks
- scalability
- token costs
- maintainability
- architectural complexity
- and other things
At the same time, many developers/content creators who heavily promoted Bubble in the past also pivoted hard into AI focused content, services, and courses.
So the message shifted from "Use Bubble to build this” to “Build everything with AI”.
More recently, we also started seeing “Bubble to code migration specialists” everywhere. And while migrations are sometimes justified, in many cases the real problem wasn’t Bubble itself, but poor architecture, technical debt, lack of optimization, and weak engineering practices. Migrating bad architecture to code usually just migrates the same problems.
And that brings us to the fourth group of factors.
GROUP 4 — Market Saturation and Low Quality Work
Over the years, the Bubble ecosystem exploded with new developers.
Suddenly:
- everyone became a “Bubble Expert”
- everyone became “Senior”
- everyone had “6+ years experience”
- everyone became “Certified”
- everyone created “over 200 apps” and “helped companies raise millions”
At the same time, many agencies started scaling using very cheap labor, often hiring inexperienced developers to maximize output volume at lower cost. That created a huge amount of poorly built projects. And many clients later had to rebuild their apps entirely or some parts, sometimes multiple times, hiring other agencies or more experienced developers.
Freelancer platforms also became saturated with developers charging extremely low hourly rates while simultaneously marketing themselves as experts. Of course everyone needs to start somewhere. But calling yourself a senior expert while charging $5/hour creates obvious contradictions.
And honestly, that alone should already be a major red flag for clients. But as we all know, many people prioritize cost above everything else, and only realize the consequences later, usually paying far more in the long run because the wrong people were hired or the project wasn’t built properly from the beginning.
Those poorly built projects almost certainly created negative experiences for many clients. And naturally, many of those people probably shared their frustrations and disappointments with friends, business partners, founders in their network, communities, and social media.
And realistically, many of them likely didn’t conclude:
“We hired the wrong agency or developer.”
Instead, they may have concluded:
“Bubble, the tool, was the problem.”
And honestly, I wouldn’t even be surprised if some agencies or freelancers themselves ended up blaming the platform in order to justify poor work, bad architecture, missed deadlines, or low-quality delivery. People can become surprisingly creative when trying to avoid responsibility.
That inevitably damages Bubble’s reputation. But to Bubble’s credit, over the years, Bubble tried to help the market choose good developers.
-
They created a recognized and certified agency ecosystem.
– Although the metrics and criteria used to evaluate agencies did not always seem to reflect actual technical quality, architecture standards, or long-term maintainability. -
They introduced developer certifications.
– Although certifications mostly validate platform familiarity, basic logic, and understanding of Bubble itself, not necessarily the ability to architect scalable, maintainable, secure, and real production grade applications. -
They allowed some developers to be listed publicly as official consultants and educators through Bubble’s own channels.
– Although those opportunities appeared to remain limited to a relatively small group over time, without becoming something more broadly accessible to the wider developer community. -
They also promoted and supported various bootcamps and learning initiatives over the years.
– And to be fair, those genuinely helped many new developers learn the fundamentals of the platform and get started much faster than they otherwise would have. -
And in the last 2 years Bubble also started investing more heavily in structured educational content around topics like design, responsiveness, security, integrations, mobile development, and related areas. And most recently some other free courses.
– Which is genuinely valuable. But realistically, the number of people who fully consume and deeply apply that kind of educational content is usually much smaller than the total number of new developers entering the ecosystem.
Those efforts absolutely help. But I think all of these factors combined, and probably others not listed here, contributed to a noticeable decline in interest around Bubble over the past years/months. Not only from founders and clients, but also from potential new developers deciding what platform to learn. Ironically, this also conflicts with Bubble’s current goal of attracting more new users, which likely explains the strong focus on AI onboarding and mobile experiences.
One thing that genuinely brought excitement back to many experienced Bubble developers was Buildprint. For many people, it felt like a glimpse of what Bubble’s future could become, at least in terms of AI. But as George himself has said multiple times, Buildprint is not really beginner-friendly. You still need at least a decent understanding of how Bubble works to extract real value from it.
And that makes total sense with another phrase I heard from George. If a newbie starts using Bubble, without any prior knowledge of Bubble, and uses the native AI to create something, they’re “vibe coding” in Bubble, just like they would be in any other tool.
And just to be clear, this is not a complaint, nor an attack on Bubble. I genuinely love building with Bubble And I try to help improve by giving feedback and suggestions. It changed my life in many ways. Honestly, building with Bubble barely feels like work to me sometimes, I lose track of time doing it.
The platform absolutely has flaws and limitations, but I also deeply appreciate the amount of work and effort the Bubble team and the community have done over the years to get the ecosystem where it is today.
So I’m simply sharing my perspective based on everything I’ve seen during the time I used the platform, and I’m genuinely curious how others see the current situation as well.
What are your thoughts on all of this?
) but there’s a lack of good tutorial, topic or useful stuff that could explain: why it’s bad, why it’s in this state, what should have been done and what could be done to fix that since it’s was built like this. Vibe coding it’s just not an “AI thing”… a lot of apps started with a simple idea and got more and more features added that bring often tech debt. Missing a long time view about the product will for sure lead to this.
I have been using Bubble since 2016, this is the biggest shift I have seen.