Announcing Buildprint AI: The agentic engineering and observability platform for Bubble 🎉

[randomanon]

@asked111

I wonder if there is even any point in bubble doing their own ai

Not if they’re gonna only allow me to use one model.

@asked111

They would HAVE to put George and his team in charge of entire AI department

Not sure why they invested in Flusk but not this despite security being far less important from a growth perspective. Maybe George is from the wrong country in Europe?

[georgecollier]

@asked111

This is absolutely game changing as means i can view/fix/edit/oversee and plan features on my app from anywhere, on my phone, at a cafe, on vacation, not just the office or laptop. Will save a lot of time.

And this is the future I’m building for.

I think in 12 months, serious Bubble developers will be making more edits using Buildprint than themselves in the editor or with Bubble AI. I hope Bubble AI comes back though.

[georgecollier]

@Newed

@georgecollier, did you make sure to prevent similar things from happening

Yup, Buildprint generally knows what actions/elements/events can accept what expressions.

Of course, it’s a non deterministic AI, so naturally it can make mistakes. Bubble’s agent does too so it’s nothing new, I suppose.

[georgecollier]

@asked111

Infact buildprint can actually HELP identify them. For example my app had a load of workflows that were pointing at folders that no longer existed. Teeny bubble bugs sitting there unnoticed…”probably" harmless…

On this:

Ask Buildprint to identify all actions that schedule API workflows with parameters no longer in the API workflow.

Suppose you have a backend workflow with Parameter Q, and you set Q to Do a search for Users when scheduling.

If you later delete Parameter Q, the Do as search for Users will continue to be evaluated, even though you can’t see the expression! This can cost unnecessary workload units and is almost impossible to detect.

Same applies for API Connector parameters that you remove - any previous expressions in those parameters will generally still have their data evaluated.

[boston85719]

@georgecollier

I think in 12 months, serious Bubble developers will be making more edits using Buildprint than themselves in the editor or with Bubble AI.

I think the sentiment is correct, however, like most AI businesses, what’s the moat for buildprint? Is there a secret sauce that can not be easily replicated in weeks or few months by competitors?

It’s likely the first of many, similar to the first chrome extension. It’s impressive, and then others replicate.

It’s pretty spot on to predict though that developers would use something like this. I was saying in past bubble should open the APIs and rewrite terms and conditions plus create a new thought that it’s not the editor people are paying for, but the infrastructure that allows apps created and hosted. The editor is mostly UI for basically an API provider.

I watched a very interesting YouTube last night about the way SaaS will move to be less focused in their UIs and more focused on their APIs, and I think you’ve demonstrated well for Bubble they should be shifting in a similar direction, and hopefully that stops the incessant UI tweaks they’ve been making and frees them up to focus on real features and bug fixes.

[georgecollier]

@boston85719

what’s the moat for buildprint? Is there a secret sauce that can not be easily replicated in weeks or few months by competitors?

1. I just don’t care about anything except building the best tool possible. I can move fast and break stuff and thats okay because its a free product and there are no expectations of it yet.
2. Execution - plenty of other logs/observability/AI tools exist for Bubble, people aren’t using them for a reason
3. I do it because it’s fun not because it’s gonna make a ton of money, so naturally that love feeds into the product. The challenge is the fun part. I don’t need a moat, we build cool stuff → community likes us → we get good reputation and work because good stuff comes back to those who contribute positively. Not everything is a direct money making product (NQU Secure wasn’t, and we subsidised like $1,000/mo in Buildprint v1 AI tokens until releasing Buildprint v2 etc)

You can see the MCP server and how it works, it’s public for everyone. That literally gives away all the technical goodies. So, if people wanted to, they could clone it, but they’re gonna have to have a good product experience to go alongside it.

Otherwise it’s a cheap knock off

also see Announcing Buildprint AI: Advanced AI, logs, and observability for Bubble 🎉 - #69 by georgecollier

[hobum91]

Wow.. I don’t have any knowledge about programming language but I can feel this is amazing work.

May I ask you something, George?

I think eople like me(who has little knowledge about code) need a tool which translate ‘good design html code’ into bubble component.

Although there are bubble official plugin (figma to bubble) and Bubble AI agent, when I need to develop user experience and UI, the googleAIstudio or lovable gives better layout and components and much more faster then using Bubble AI agent.

I often get frustrated because redrawing pages made by code(e.g. the page made by lovable) into bubble(e.g. a kind of handmade working in bubble editor) spends several days..

The tool which can draw groups and buttons into bubble editor by pasting HTML or CSS code is needed desperately….

Is there any way you could save us..?

[georgecollier]

@hobum91

The tool which can draw groups and buttons into bubble editor by pasting HTML or CSS code is needed desperately….

Is there any way you could save us..?

Yeah it’ll be able to do this very well. Sign up, send me your email and I’ll grant you access and you can try it.

[gf_wolfer]

@georgecollier finally getting around to testing this out. Love the work you are doing for the community, better tooling for advanced users is so important for those of us with ‘legacy’ apps

[arthurdukes]

Hey @georgecollier, would love to jump into the beta and test this out too? Is it compatible with bubbles native mobile apps too?

[code-escapee]

@boston85719

like most AI businesses, what’s the moat for buildprint? Is there a secret sauce that can not be easily replicated in weeks or few months by competitors?

I don’t think Buildprint is trying to build some massive moat. From everything George has shared, this feels much more like building tools to genuinely help the community.

But if we’re talking moats, the more interesting question isn’t Buildprint’s.

It’s Bubble’s.

With AI in the mix, what actually stops someone from building a focused alternative that includes all the things many of us have wanted for years?

• Better native UI components
• True parallel processing
• Clean loop primitives
• First-class backend logic
• Native AI integration from day one
• Proper backend ↔ frontend realtime communication
• Fewer cosmetic UI tweaks, more execution depth

And instead of optimizing for the 99.47% of apps that never monetize, you design specifically for the serious building complex, revenue-generating apps (not the landing pages).

Think complex F&A platforms. Internal tools that actually run businesses. Regulated workflows. Real AI apps…

What does that take today?

1. A small team of crack devs (with strong dev ops)
2. And honestly… a handful of Claude Max subscriptions.

The lessons are already learned. The wishlist is already written (throw in the top 100 used plugins to boot).

If you stay focused, it’s not a five-year platform play. It’s a matter of months to get something serious off the ground.

And without investors demanding hockey-stick user growth, a platform built for those real builders could be very profitable. Just saying…

[senecadatabase]

@code-escapee

But if we’re talking moats,

I understand what you’re saying about this in a business sense, but you’re over simplifiying it.

A ‘moat’ in the business sense doesn’t mean someone can’t copy it…

we could have someone copy Bubble. Would they be as successful? No. Not unless they had a whole lot of money behind them.

We can all copy apps…but, are we going to be as successful? No, probably not.

There’s a whole science behind the whole ‘moat’ thing that I don’t think you’re taking into consideration

[randomanon]

@boston85719

I think the sentiment is correct, however, like most AI businesses, what’s the moat for buildprint?

Distribution and more importantly, trust. People aren’t going to be uploading their proprietary business logic, data, etc. into any random product. George has shown that he cares about security. Also, look up what a “funnel” is.

Finally, when the company doesn’t care enough about its own platform to prevent it from going under, and you have invested in said company, it’s aligned with your interests to help prevent it from going under.

[boston85719]

@randomanon

Also, look up what a “funnel” is.

That’s buildprint. It’s a great funnel, and gets at what I said not long ago. Soon enough the initial build in bubble will be free or very cheap, in order to get the customer for ongoing maintenance, iterations and new feature adds.

@code-escapee

With AI in the mix, what actually stops someone from building a focused alternative that includes all the things many of us have wanted for years?

Somebody built alternatives before AI…do you know the names of them?

@georgecollier

free product and there are no expectations of it yet.

Awesome. If the moat is it’s free, then yes, that disincentives other agencies from releasing their in house AI builders publicly and likely buildprint is the easy choice.

@georgecollier

we get good reputation and work because good stuff comes back to those who contribute positively. Not everything is a direct money making product (NQU Secure wasn’t, and we subsidised like $1,000/mo in Buildprint v1 AI tokens until releasing Buildprint v2 etc)

Yes, not a direct money making, but built for purpose. I can see how if those two earlier releases increased revenues for NQU, while being free, it’s a solid idea to expand the client funnel product range.

I can imagine if it’s for the love and challenge, you’ll be quite busy with it for a while. Also a great way to speed up internal processes and get increase in effective hourly rates. Genius stuff and impressive.

[georgecollier]

@randomanon

when the company doesn’t care enough about its own platform to prevent it from going under, and you have invested in said company, it’s aligned with your interests to help prevent it from going under.

@boston85719

built for purpose. I can see how if those two earlier releases increased revenues for NQU, while being free, it’s a solid idea to expand the client funnel product range.

Indeed! Our sales funnels are some people’s entire SaaS hahaha

Will it become a paid product at some point? Sure, but it’ll be priced so that people are happy to pay given the value they’re getting out of it and people just using it occasionally can use it for free…

[hussain]

Hey George would love an invite when you get a second. Let me know if you need more info.

[parthapatel0723]

Just saw this thread after a while, and wow, you’re just cracked @georgecollier. This is honestly just crazy. Are you eventually planning on opening webhooks for Buildprint so new logs and stuff can get pushed to my logs platform?

[eple0030]

@eple0030

Hey George, i would love beta access too!
thanks a lot!

Etienne

[georgecollier]

Log citations

Buildprint can now cite logs directly in chat. This lets you easily understand and investigate real production issues.

Files and attachments show in chat

Any files created by Buildprint (e.g Markdown docs) are available in chat and will open in a sidebar for preview.

Fixes and improvements

• Fixed issues with performance on long chats
• Draft messages now persist between conversations
• Review comments sorted by severity
• Review creator uses default model preference
• Improved API Connector editing
• Improved element and UI design
• Upgraded to Sonnet 4.6
• Improvements to logs explorer speed

Editing apps with Buildprint

This week, we released Buildprint editing to a small number of users. It’s been incredible to see the response from these users, which range from non-technical users who built their own app, to power users and professional developers working on some of the largest apps in the ecosystem.

Some highlights:

• one team used Buildprint to organise 340+ uncategorized workflows into folders
• a solo founder used Buildprint to overhaul the design of their marketing pages
• a marketplace owner used Buildprint to identify and fix bugs with payment issues
• a new user used Buildprint to set up all of the Stripe API calls they needed for their app automatically

Buildprint editing is going to be available on all apps, but is designed primarily for complex existing apps. Buildprint is not going to be a ‘vibe coding’ tool. We are building this for serious businesses that are interested in releasing good quality software on Bubble.

You can read more about the editing beta and what’s in store here.

[code-escapee]

@boston85719

Somebody built alternatives before AI…do you know the names of them?

Not following? Do you mean that they never became a household name?