Monthly Community Update -- August 2025

Announcements
1

Hi everyone,

This is the August community update. You can read last month’s update here.

I’ll talk more about this below, but with the AI app generator out of beta, we’re making fast progress toward conversational editing. We think vibe coding is on to something, but why are they outputting JavaScript? Most people using these tools don’t know how to code, and getting the AI to do what you want without creating bugs leaves you stuck in a prompt loop trying to fix code you don’t understand. Our approach, which outputs in Bubble’s no-code language instead, is visual and easier to understand. You can give the AI better instructions and there’s less room for bugs in the output. It also comes with batteries included: pre-built authorization, security, database, and hosting. All that to say: We are heads down working to make vibe no-coding the industry standard.

On to the updates!

What we shipped last month

Like I mentioned, the AI app generator is officially out of beta! We’ve paired this with a few product enhancements, including updating the prompt box to include starter prompts for apps we see a lot of users creating. Try it out here.

We also improved the way dynamic expressions are displayed in the editor after an app has been generated by AI. Before, it showed raw expressions; now it shows the typical placeholders you’d see if you built the app from scratch yourself. This makes jumping into the editor after generation more intuitive for new users.

Alongside that release, we just shipped v0 of AI data type generation. Describe the feature you want to build, and Bubble AI will set up the data types and privacy rules you need. If your plan has changelog access, you’ll also be able to see a record of the changes that Bubble AI made when you used the feature. This iteration is best for new apps and users getting acquainted with database set-up. Next, we’re working on functionality that would allow Bubble AI to reference and edit existing data types. Access it via the data tab in the editor. You can read more about what’s on our roadmap on the blog here.

Build guides, which help you map out features for your app idea, got an upgrade from GPT 3.5 Turbo to GPT-4o mini, which means faster response times and more reliable answers. You can access build guides from the sidebar on the homepage when you’re logged in.

Outside of AI work, our new Figma to Bubble conversion tool is live! This was originally a community-built plugin known as Deezign, and now it’s Bubble’s second-ever acquisition. It’s been exciting to work with @Bek to integrate it into Bubble officially and make it free for everyone. (Before, it cost $25/month.) The converter allows you to create beautiful front-ends and is especially useful for design-first builders. Learn more about getting started here.

The team completed phase one of improvements to the mobile app deploy flow, which should make it more straightforward to launch in the app stores. Phase one involved removing settings beta testers flagged as confusing or unneeded, and we’ve already seen fewer build failures as a result. Phase two will involve adding a validation step for App Store developer keys to make sure you’re set up to launch properly. That work is already in progress.

We also made some changes that significantly speed up mobile app load times, especially on Android. Try deploying a new build and testing your app’s performance to experience these updates. You can read more on the forum here.

Lastly, we made some improvements to the native map element. You can now kick off a workflow when a map marker is tapped, see your current location as a blue dot on the map, and we also addressed issues with custom map markers in production.

The team officially wrapped work on migrating Bubble’s main cluster off of PLV8 (and they’re on track to finish up migration work for dedicated servers in the next few weeks). This makes our database significantly more reliable and opens the door to more performance improvements down the line. We’re also working on optimizing database indexes, which will speed up performance.

Over the past few quarters, we’ve been working to help users build as securely as possible from the start. This month, we launched two security initiatives:

  • Alerting users if their editor is publicly accessible, which could expose sensitive data or allow unauthorized changes. PSA: Making your editor publicly available should only be used for demo or open-source apps where you don’t have any secure credentials or data. We sometimes see this setting turned on for apps that look like they are meant to be for business use, which is a critical security error. We sent out a round of emails to apps with public editors, to encourage people to confirm that the app is intended to be open to the world. Just in case you missed it in your inbox, it is worth 10 seconds to confirm you have this off for your own app.

  • Removing a vulnerability that could reveal sensitive app info during the setup process for the API Connector. This makes API connections more secure from the get-go.

Last but not least, thanks to user feedback, we updated our bug reporting flow. It’s now quicker and easier to submit the impacted App ID and a description of the behavior, which is then automatically escalated to our human support team. Previously, you were asked to submit a screen recording and extra details up front. Thanks to everyone who gave us their feedback on this.

What we’re working on

  • Mobile: The team is hard at work on our main post-launch focus areas I talked about last month: in-app purchases, deep linking, and the mobile plugin editor (currently in alpha). We’ll have more to share soon. Since the beta launch, we’ve also made good progress on AI app generation for mobile apps, which should allow even more people to create for mobile even faster.

  • AI: Now that the app generator is out of beta, we have a few additional improvements we’re working on, including a feature that would convert an image (like a wireframe sketch) into an app page. The majority of the team has moved on to focusing on conversational editing, the next stage of our AI vision for Bubble. Iterating by telling the AI what to change will help you develop faster, but you’ll still maintain ultimate control over what’s changing in your app. This is critical for users building apps that can scale.

  • Property editor: We’re still conducting user testing for a revamp of the property editor. A big priority is testing with power users to make sure these updates work for you and users with less Bubble experience.

  • Marketplace: We’re working on bringing Bubble coaches into the same search flow as agencies and freelancers so that people have a one-stop shop for help, whether they’re looking to outsource their build or learn how to use Bubble themselves. That should be rolling out by the end of the month.

How to get involved

We’re opening applications for the next Bubble Ambassadors cohort soon! Keep an eye on Bubble’s social media and here in the forum for more details.

Bubble is hosting a number of events over the next few months:

If you’re interested in hosting your own event, make sure to visit our Community page to submit it to our events calendar.

Finally, there are still a few days left to enter our Summer Mobile App Challenge for a chance to win $5K. The deadline is August 18, so make sure to leave enough time to submit to the app store! (Approval through Apple or Google can take up to two weeks.) Get more details about the contest here.

New hires

We had four new hires join the team this month! Welcome to:

  • Pavitra, Vivian, and Elizabeth, software engineers

  • Jimmy, senior software engineer

That’s everything for this month!

— Josh and Emmanuel

17 Likes
3

This is great news. It’s painful to come back to the Bubble world and start clicking and typing to build things once you go to conversational efficiency. The outcomes do not have to be slop if you know how to converse with the machine. So full support from me for this :clap::clap::clap:

3 Likes
4

:partying_face:

in case anyone’s wondering if it’s some hidden thing that was fixed, this is the initialisation response data, which is now no longer included in the page’s client side code and is sanitised when initialised - so you might’ve already been aware of it but now don’t have to think about it, and existing apps get that benefit!

hyped for this

I guess this refers to the ‘canvas placeholder’ option?

2 Likes
5

@josh Can you maybe make a statement about some long standing “beta” features that either, doesn’t receive enough attention from team to fix issues or add requested features or that just seem to be completed?

A) Table element
B) Expression parenthesis (why it’s not default?)
C) Snippet on hover
D) Native support for streaming API (not Beta… but should be if we consider issue and limitations…)
E) New workflow editor (some requested features that I think was considered but not yet released…)

Probably forgetting other releases/beta…

10 Likes
6

Are there any plans to introduce a native JSON field type?

Right now, working with JSON is very limited. We often end up storing raw JSON in text fields, manually parsing strings, or relying on third-party plugins that don’t integrate cleanly with the rest of the editor. It would be incredibly useful to support native manipulation of JSON objects directly in the editor.

Curious to hear if this is something on your roadmap.

10 Likes
7

Excited to see where this leads. Great stuff as always!

8

Looking forward to this!

9

Can’t wait for this revamp to be completed, in part because:

  • Custom States will finally be given their rightful, more prominent and accessible place in the UI.
  • @georgecollier will have one less reason to hate on them. :smirking_face:
6 Likes
10

Thanks Josh, this is the kind of stuff we want more of. Security and fundamental performance improvements.

A while back it was discussed that there would be a way to obfuscate fields instead of having all of them be publicly available, is this still somewhere on the horizon? DB names and design can inadvertently leak sensitive information. At the very least if you can’t hide the schema then a “public name” and “private name” would be much appreciated.

This would be the single biggest enabler for new Bubble apps, given how most of them will have AI responses integrated into them in some way.

2 Likes
11

Tell me this is true. Apparently I’m not a power user so i don’t get invited to try new toys.

On a serious note. I wish Bubble had more actual power user tools. My wish list to Santa:

  • Better state management. I hope this comes with the new property editor.
  • More user control over CSS. Current methods don’t take advantage of browser caching.
    • Variable element widths and heights at least
6 Likes
12

@josh @fede.bubble what will the AI features being added for app generation do for keeping workload units down?

It would be cool to have AI features that analysed the app and suggested ways to minimize workload unit costs (if it’s a possibility)

This would be beneficial to Bubble and the developers!

13

Right now the team is focused on bringing conversational editing to the editor so anyone can build the front end of their apps in minutes.
After we’ve completed that milestone the AI will start bringing AI to other places on Bubble. Totally agree that a magic AI optimizer would be awesome

2 Likes
14

Sounds good - looking forward to the features you’ve mentioned :ok_hand:t2:

15

You should save them as API Objects instead.

Plugins integrate well within the editor as they are Bubble plugins and use bubble workflow actions, triggers etc. There are ways to use free plugins to manipulate the raw JSON text, but others that help with manipulating the API objects are the best ones to use as those API objects are available in dynamic expressions just like any other type of data.

This is true. I’m not sure if it will be put on the roadmap though as it is the type of thing that already has enough support to achieve without native integration. Part of the reason Bubble switched to WUs is because some apps were storing raw JSON as text fields causing unreasonably resource intensive queries.

But I do understand the point of how powerful using JSON is in a Bubble app. In my use of a Hybrid Data Structuring approach where I use API objects and store them in the database, or as a file, or just as text directly on page, has brought so much more power to my apps it is mind boggling. It also helps reduce WUs drastically. More developers should be exploring ways to get the same kind of benefits.

Yeah, it is amazing how well JSON and AI go together.

That is nice to hear the team is taking into consideration feedback from impacted users on this. I think though the approach decided upon misses the mark though. The description seems to imply the issue was pushed further down stream rather than being addressed and cleaned up. So instead of needing to deal with the AI asking these questions, slowing down the time it takes to submit a bug (the feedback for why there was an issue with the bug reporting) on the initial submission, the issue of slow down will now be amplified as it will require an back and forth via Email between the User and the multiple support agents that will be involved in the report process (usually 3-4 before somebody has it escalated high enough to be investigated).

I believe the real issue is the slow down that was caused by replacing the old working solution of a simple form for an AI bot. Instead of pushing that slow down (ie: bottleneck) further downstream in the bug reporting process, it would be optimal to remove the bottleneck. The easiest and most efficient way to do that would be to provide the bug report form again. It is simple, it already existed, just need to add the page back. Users can then use AI bot if they choose, or use a simple, fast and efficient form to submit them as we had previously.

I had requested that prior to the release of the Flexbox responsive engine during discussions with the team as to what should be added. I continued to voice that opinion on the forum just after the release, and months later. After a while, I just decided to build and release a paid plugin for that. My hope was the flexbox responsive system would allow for dynamic values in all input areas and that all layout and styling options were to be available in the conditionals tab as well.

It would most likely just have the most basic approaches implemented, but not the approaches that truly reduce WUs. There are ways to reduce WUs for searches by up to 95%, bulk creation and updates of data by 99%+ and they even speed up the load time of the content.

I agree

Awesome!

Winning

No idea what the beginning of this means, but the outcome part I get and it makes me excited.

How can we get involved in this?

3 Likes
16

This needs to be an urgent priority, bubble is obsolete in terms of search performance and data recording.

2 Likes
17

you make me feel confident. i’m looking at the things u talk about the most, and in which order. i like it.

nice work @josh

also - shout out to @nick.carroll’s work on native. i’m liking it so far.

2 Likes
18

Hope we can get a clear timeline on using custom SDK’s for mobile @josh Thanks for all the hard work you and the team put in.

19

I think it’s already possible but access to plugin editor is actually restricted to some users.

1 Like
20

Ideally, Bubble’s map element would let you send a marker’s data to another page or element. For example, tapping a Starbucks marker could navigate to its store detail page and display the address, hours, menu items, and any other fields you need.

Right now, the map element doesn’t support passing marker data the way a repeating group does.