Are Bubble devs doomed?

Meta
1

I’ve been thinking about these questions a lot lately. Do Bubble skills become redundant within the next 12 - 24 months? Can Bubble meaningfully compete with AI coding agents?
What is Bubble’s unique value proposition 12 months from now? Should I be exploring alternative career options and if so how aggressively should I do so?

The umbrella question becomes, do AI agents really replace us all?

My opinion is initially YES, but eventually NO. Please feel free to agree or disagree with me in the comments of this post as I state my case.

The current global economic model that all emerging and leading economies subscribe to is founded in the basic principle of globalization and economic specialization.

Globalization can be understood through the lens of specialization, historical advantages, and the random displacement of natural resources. These factors foster economic interdependence among nations, where countries focus on producing goods or services in which they have a comparative advantage. These advantages often stem from historical developments, geographical positioning, or the availability of specific natural resources.

Now! Does the AI agent model fit into the current global economic structure where countries produce things that they have a comparative advantage in? Before we answer that question, lets try and dissect what exactly qualifies as an AI agent, what is their unique value proposition and how does the AI agent model affect non-AI specialized nations (pretty much every country but the United States and potentially China).

I asked Copilot what an AI agent is and here is what it said: "An AI agent is a computer program or system designed to perceive its environment, process information, and take actions to achieve specific goals." Sounds a lot like what humans do in the workplace right? but wait … there’s more.

Copilot: “AI agents excel in specific areas, especially where speed, accuracy, or handling large amounts of data are involved.” Copilot failed to mention that at some point AI agents will develop better reasoning abilities than the average human worker.

So in a world where AI agents are faster, smarter and less prone to human error than human workers … how do people and by extension Bubble devs compete?

Well before we try and answer that question, lets introduce a concept I call AI agent maturity. I use this term to loosely define AI agents that can economically complete highly complex tasks without human intervention. Other people might call this AGI.

When AI agents reach maturity, people and by extension Bubble devs cannot compete at all … at first!

This brings me back to my point where I very boldly stated that people and (once again by extension Bubble devs) get their jobs back. How does this happen? Well to explain how this happens lets draft a little case study based on a singular country, Germany.

Germany has approximately 1 million software developers source. For the sake of this case study, let us assume that the average software developer in Germany earns $40 000/year and pays 30% income tax to the German government. If this is true then the German government collects 0.3x40000x1000000 ($) in tax revenue per year. This equates to 12 billion usd in tax revenue collected annually from German software developers.

Now assuming that AI agents are fundamentally human replacement software and AI Agent maturity is achieved, 1 million German devs would at least temporarily be displaced and the 12 billion usd in market value would be injected into AI superpowers (ceteris paribus) namely the United States and China.

This case study makes a lot of assumptions, many of them not theoretically sound. However, this case studies aims to pursuade that the AI agent model doesn’t fit into the current global economic structure as the only significant winners would be AI superpowers. Governments of non-AI specialized countries would have an obvious incentive to collectively ban human replacement technology at a large enough scale. And that is where we get when we get our jobs back.

This is purely a speculative article inspired by the current global economic structure. Please feel free to share your thoughts.

[I just so happen to be a front-end dev who’s open to work. Consider shooting me a direct message if you have an open position.]

Looking forward to all your thoughts on whether we should be worried as Bubble devs or not and why?

4 Likes
2

There’s a lot in there and correct me if I’m wrong but your question seems to be:

“If AI agents get better in the future and the amount of people learning to code continues to increase, why should I spend my time with Bubble?”

In general, you should consider that Bubble will also benefit from AI innovations. Better, faster, cheaper models can be integrated into Bubble’s ecosystem (that what the team is working on with Bubble AI right now). But purely AI-based agentic app building will always fall short of the controllability and customization a visual editor like ours gets you.

Also, I don’t think this will be a single-winner market. Most likely Bubble and other tools will serve different purposes and will fit the needs of different users. Of course, at Bubble we think we should fit the needs of everyone, and that’s why our roadmap includes AI but also Mobile (react native) development, Editor improvements, backend stability upgrades, etc.

Hope that helps!

9 Likes
3

I wouldn’t underestimate the amount of people who enjoy working with a person and who would reach out to a developer based on their own inability to develop or use AI to develop.

Using AI is a separate skill. Using AI to develop is a separate skill.

I wouldn’t write your skills off. Exceptional ability in Bubble is a valuable skill, because there are a lot of mediocre users of Bubble.

Bottom line: you don’t have to wait for the whole full circle you described-- Bubble is relevant today and I believe it will continue to be. Because while a lot of people can code, those people are expensive and slow, even working with AI.

1 Like
5

I would say Bubble devs who do not have that inner motivation to be the best at what they do are doomed. I also would say that Bubble devs who do not have that inner motivation to learn how to work better/smarter/faster with AI are doomed.

What I have seen happening is a lot less questions being asked on the Forum by newer users…that is despite the fact that the introduction of the AI for Bubble has causes a 1.5x increase in retention rate…so the AI is keeping new users around Bubble longer, but it is also, seemingly making it so that less of those new users are really learning Bubble and so do not have questions to ask on the forum.

I imagine most of those users will at some point find the AI falls short and they will just walk away completely. Others may have that will and determination to really see their dream come true and are a true blood DIYer and they will then decide to really learn Bubble to finish their project.

I’d say then there is the group of new users who are not even actually interested in attempting to use AI to build their app, but play with the AI builder, see what it can do as a starting point, assume that makes all Bubble developers LESS VALUABLE and start to make their way toward the cheapest developers they can find. Because what happens due to the AI, is that no longer are you a Bubble developer, now you are just somebody who uses AI to build apps. So you skillset as a Bubble developer just took a major nosedive in terms of its value, because it will be perceived as not so unique since most people will just assume all you do is use the AI.

Other new users will have a project that they already personally deem as too complex or too important to trust to an AI or an intermediate level Bubble dev. These users will likely continue to understand and value the skill set that true Bubble experts possess.

Ultimately though, it is not the AI itself that will cause all of this…it is going to be the host of other knock on effects from the AI and the new products/services that truly experienced Bubble devs will be introducing that will make it even harder for the intermediate Bubble dev to express and communicate their value to potential customers.

If as a Bubble dev you can confidently say that you know more about how to build on Bubble than the AI does, you will likely be okay, but must start making the necessary adjustments to the client funnel and the way in which you work with clients and AI.

When AI first hit my radar and Bubble begin chattering about it, I was nervous, now when I see what it is doing and understand AI more in a general sense of it’s capabilities and limitations, I’m excited for it. It has already completely changed the way I build in Bubble and operate my business and I have not even used the built in Bubble AI yet. In fact, I’ve been on the fence for 7 years about ‘going all in’ on Bubble and starting an agency, but now, ‘I’m all in’.

8 Likes
6

There are always people who will pay to have something done.

We have coffee pots we can control with our smartphones. They grind the coffee, keep it at a constant temperature, and even froth milk.

Yet, many people go to Starbucks every day and pay $8 for a cup of coffee.

It’s also like asking if AI will put mechanics out of business because AI could probably tell someone what’s wrong with their car. Well, it probably can tell them, but, then they’d have to get their hands dirty so they take it to a mechanic.

Oil changes aren’t that complicated…yet the oil change industry is a $160 billion dollar industry.

People can clean their homes…but some pay to have it done.

We can detail our car…but some would rather pay to have it done.

People pay to have their dogs walked and groomed…when actually they can do it themselves.

We can cook at home, but would rather go out and spend a couple hundred dollars at a restaurant sometimes.

A lot of businesses are too busy to do a lot of things so they’ll gladly pay someone that’s good to do it for them.

The list goes on and on and on.

I’ve said before, that I think Bubble has an advantage with the AI scene because they also offer an editor. I’m looking forward to seeing how Bubble progresses with its AI.

But anyway, no, I don’t think AI will have a significant impact on Bubble Developers as there will always be people who will pay to have something done for them.

8 Likes
7

Valid points and the fact that you ask them and others are wondering the same tells the whole story. Very few people really understand what is happening at the moment. AI will eventually almost replace everything.

If AI can build anything you want in seconds, on a dollar per month hosting plan, why would you pay bubble 10 or 100x to either 1. Get exactly the same as what AI delivers. 2, get to your goal less quickly because you have to deal with Bubble specifics.

I do think that the SaaS market will explode but more as microservices doing either one task very well or integrate many tasks in a valuable way. But this will be done at 10% of current SaaS costs and companies of one or a few matter subject employees. Nothing more or less.

If you are really interested you should lookup Williamson about transaction cost economics and Coas about the theory of the firm. Eventually, AI will drive transaction costs close to zero meaning large organizations are not needed anymore. Also, a few companies will dominate the world. Companies they deliver the AI models like Open AI, Anthropic, Google etc. This will continue until we live in a world controlled by machines.

8

There are 30,000+ live Bubble apps that need developers and won’t disappear overnight.

Bubble skills (well, the way of thinking that it teaches you) are transferrable to ‘AI code’ and other low-code platforms.

3 Likes
9

“A new technology has been developed that will replace the old technology.”

—

Meaning while in 2025

image
image1290×1871 282 KB

6 Likes
11

Software Developers - whos tasks are to model business processes and abstract business logic into Software - now have new AI tools to use. And they are powerful tools that makes Software creation easier.

If your Software Development tasks are typing code and configuring the setup of systems to someones specification - I’m afraid those tasks are now obsolete.

If your tasks are unpicking and making sense of some messy design and convoluted logic - there will be plenty of work for you. :slight_smile: and the new tools will help.

5 Likes
12

I think competition in every market is about to speed up rapidly. Everyone is focusing on coders at the moment because it is one of the obvious use cases for AI, but in 5 - 10 years it’s going to be almost all industries and jobs.

In my opinion, in almost every industry we are going to see dramatic increases in efficiency, making one person able to handle the workload of what would take a team of 10 in 2024. This will mean that the top 20% of employee’s in every sector will still be needed, but I worry for the people that aren’t passionate about their craft and just “go with the flow”, because theirs a good chance AI could replace them.

Regarding Bubble’s future, I still think the backend tools and NoCode infrastructure that Bubble offers is the best in the market. Lovable will build you a nicer front-end and get you from 0 to 0.6 quicker, but if it’s a race from 0 to 1 I still back Bubble.

Could that change in the next 12 to 24 months? It’s very possible, but we will have to wait and see. Until someone can make the NoCode Backend functionality that bubble has, I don’t think so.

Different use cases will also be important. I think MVP development and SaaS will start to lean more towards lovable (Because it’s open source and you can validate ideas very quickly and cheaply), but I would argue internal CRM’s and Tools are still better built on bubble because of the backend complexity and they tend to not scale past a certain number of users

Connect with me

13

We are building with AI only for 6 months now. I am curious to learn from you what you think is not possible in the backend that bubble handles now as no code tool

14

A short story

IMG_0044
IMG_0044828×1098 143 KB

IMG_0045
IMG_0045828×680 73 KB

IMG_0046
IMG_0046828×1086 130 KB

13 Likes
15

Thirty years ago, with the rise of personal computers, many predicted that paper documents would soon disappear. Yet, in some sectors, their use has actually increased. Similarly, despite advances in AI, the number of IT job openings continues to grow, and so does the global population—meaning the demand for services keeps rising. At the same time, the overall quality of education and workforce capability is declining. The world will always need professionals who can understand, create, and manage complex systems beyond what AI can automate.

1 Like
16

I think this is a really good point. However, with the amount of money being poured into AI particularly by US Venture Capital I would lean towards the opinion that everything that can be automated will be automated eventually. The ultimate goal for AI agents will most likely be completion of highly complex tasks with minimal human intervention. For example recovering $28 billion in investment by Microsoft into ClosedAI (aka OpenAI) will require signifcant market share & significant profit margins per user. Non-automated task means money to an entity other than Openai. Naturally Openai has an incentive to prevent that from happening.

17

This is rough. Do you have more information about this? I’d assume this guy didn’t have a great understanding of basic privacy rules although I don’t know for sure since I haven’t used Cursor. Also what incentive did bad-actors have to do this? Also from this post I can’t really decipher whether the platform Cursor generally has security issues or whether the platform allows its users to create unsecure apps.

18

Not 100% sure I fully understand the question you’re asking, I will assume you’re asking what tasks cannot be completed on Bubble’s backend right now and potentially continue to be the case in the future.

My general understanding of this is that because Bubble requires work arounds and not so obvious custom solutions for complicated applications (particularly on the backend) AI will not be a viable solution for these kinds of problems simply because AI is prone to make mistakes. From my experience in backend development, Bubble requires 100% accuracy else your backend workflow simply won’t work as it should. Backend workflows are often connected to other Backend workflows. So a ‘non-accurate’ backend workflow may potentially prevent other ‘dependent’ workflows from triggering or for example may not return an API call with a really important parameter used in other connected subsequent backend workflows.

Why is this not as a big a problem on the front-end? Well its probably because backend workflows are primarily focused on data manipulations. So creating, editing and deleting data to be displayed or used in front-end logic.

So while a lack of accuracy on the front-end is still problematic, it isn’t nearly as problematic as a single error in the backend. This is further exacerbated by the lack of detailed error reports in a backend workflow that doesn’t work as it should, which makes finding and fixing an issue particularly by a relatively new Bubble dev … so much harder!

19

I agree with your logic here to some extent. However, you’re assuming that ‘Debugger Agents’ either will not exist or won’t be as good as an experienced ‘Human Debugger.’

So I’m somewhat hesitant to take peace of mind in this :slightly_smiling_face:

20

It was to a response from someone else that Bubble remains needed.

What we have done is structured the highways, road signs and traffic rules in such a way that every piece of code is consistent. Therefore we get less and less errors although our application grows in complexity.

Doing it like this allows us to have complex frontend and backend well connected, performant, secure, extendable and well documented.

It did take us 6 months though with learning every single day, rewriting pieces and more.

We are getting close to a state where we can decide to allow others to use it as well but it will be nothing like NoCode. More like just prompting and getting questions as to what users want. When ready and agreed on requirements it can build what the user wants 80% is ready and more than enough for 80% of customers, the other 20% can extend what is there by AI or add custom code. However, this last step will cost at least another 6-12 months and therefore also money so not sure if it is the best business decision.

21

Don’t you think this might be a temporary market correction for primarily driven by capital abundant opportunists and that they may be a steep draw back once AI agents become more ‘capable’

22

I agree with the premise of your argument here.

I also agree with the point you made about some people valuing convenience more than other factors such as cost.

However, let me draw up a speculative scenario.

Imagine AI agents achieve the following at some point in the future:
Average time taken to …

  • build a ‘complicated’ application: <5 minutes
  • conduct a rigorous app audit based on project scope & modern security standards: <5 minutes

Average cost of …

  • App development: 10 - 30% of normal market cost
  • App audit: 10 - 30% of normal market cost

Customizability:
-Ability to draw up 5 or more mockups of an application that meets the project scope, passes the app audit, different front-end & backend designs which have different performance implications. Imagine all 5 mock ups can be released to different batches of users which would essentially enable the General Project Owner to analyse the data and draw conclusions on the mock up that has the best performing market fit.

This is a hugely speculative scenario unlikely to be achievable within at least the next 2 years, but the point im trying to demonstrate here is that when there are ‘insignificant’ difference in a set of choices weighting convenience or personal preference over other factors is an easy choice to make. However, when there are ‘significant’ differences among a set of choices which in this context is development time, cost and customizability, personal preference may seem like a simply bad decision.

I’ll give you one more example. Imagine a Person 1 wants to travel from point A to point B. Point A and Point B are 100 km apart. Person 1’s personal preference is to horse-ride from Point A to Point B. If Person 1 travels by car they will reach Point B in 1 hour, however, should they choose to travel by horse they will reach Point B in 2 days. Lets assume that the cost of travel per hour on items like food and water is $5/hour. If person choose to go with their personal preference they will pay 48 times more over the course of the trip than if they chose to drive a car to Point B.

So in conclusion, I think your argument stands if there isn’t a significant difference among a set of choices. However, if there is a significant difference among a set of choices, the end result/decision may differ.

But that’s just my personal opinion :sweat_smile: