Bubble co-founder and co-CEO Emmanuel Straschnov opened August's livestream with demos, and the agenda was full this month. Much of it came from Bubble's scaling team, which has spent recent months on improvements for large applications, large teams, and heavy traffic.
The rest of the session covered new additions to Bubble's expression language, sign in with Apple for native mobile apps, and about half an hour of community questions on the Agent, page performance, mobile, and Bubble's 14th birthday. Emmanuel closed by announcing the winners of the Bubble Free-For-All Challenge, our most recent hackathon with Contra.
Watch the replay below, or keep reading for a summary of everything he covered.
Demos, part 1: Scaling work for large apps and teams
Emmanuel started the demos with Bubble's recent scaling work. Large customers often come with large applications, large teams, or heavy traffic, and in practice those three tend to arrive together.
The four updates he covered here are all aimed at customers on dedicated and Enterprise plans. But the value isn't limited to them, since these changes also show what scaling on Bubble looks like for apps still growing toward that size.
Read replicas give heavy queries their own database (2:26)
Customers on dedicated plans have been able to query their app's database directly for a while now. The catch is that a complex query competes with the live app for the same resources, so reporting or analytics work could slow down production.
Read replicas separate the two. Bubble clones the database into a second copy that mirrors every change from live in real time, and heavier queries can point there instead. That covers complex SQL, more advanced analytics, and feeding app data into an LLM for insights or reporting, all without adding load to production.
Before this, Emmanuel says, "there was a little bit of a trade off in terms of how far you can go and how stable your platform is to your users."
Read replicas are in private beta now, with a rollout to more dedicated customers over the coming month.
The Enterprise Hub gives dedicated customers a view of their own infrastructure (4:44)
The Enterprise Hub, which launched the day of the AMA, is a new dashboard for customers on dedicated environments. Emmanuel demoed it live using one of Bubble's own instances. Top-level status sits at the top, with more detail further down the page:
- Status and uptime shows public domain availability alongside underlying server uptime, so there's no need to ask Bubble's team to pull it.
- Infrastructure health covers metrics like application CPU, database CPU, and available memory.
- Reporting and usage highlights which pages are getting the most traffic, which connects infrastructure behavior back to actual usage.
Previously, a dedicated customer wanting this information would ask the technical success team to look into it on their behalf. Making it customer-facing means both sides are working from the same data, and access is a one-click login with a Bubble account.
Merges are faster, and they no longer block the whole app (6:47)
Merging has been a major workstream over the past eight weeks, with an entire team on the algorithm and the merging engine. The problem shows up on large apps where several people work across branches. A server-side merge used to hold the entire app locked for its full duration, and touching the app before it finished could produce odd behavior.
Now the lock is scoped to only the part of the merge that needs it, so unrelated branches stay editable. Emmanuel demoed this by merging two branches while editing a third in another tab (8:05).
Speed improved alongside reliability. Merge time is down about 30% in internal testing, since the engine now skips sections neither branch touched instead of comparing everything. False conflicts on large workflow expressions are fixed too, as are favicon differences registering as conflicts.
Emmanuel called this the most significant improvement to the version control engine since Bubble initially shipped it three years ago.
The editor and deploys are noticeably faster on big apps (10:27)
Editor load time is about 11% faster on average. "What we're optimizing for is really the slowest cases," Emmanuel says, because averages are weighted by the many small apps that already load instantly. Some of those slow cases dropped from around eight seconds to three.
Deploys followed the same pattern. The average improvement is 10 to 11%, and on some larger applications the team cut deploy time by 70%. Both sets of changes shipped across several releases over the past eight weeks, with more in progress.
Demos, part 2: What’s new in the editor
From there Emmanuel moved into changes anyone in the editor will notice. Most of them continue the expression language work he demoed in July, aimed at making logic and data expressions less cumbersome to build.
A toggle operator, and a cleaner way to convert to dates (13:17)
The new :toggle operator takes a list and an entry of the same type, then removes that entry if it's already in the list or adds it if it isn't. Text, custom fields, and custom types all work. Doing this used to take a couple of workflow actions and a condition to test whether the item was there. Now it's one operator inside one expression.
There's also a new expression to convert an entry into a date, shipped over the past three weeks. Both changes come from the same pattern. Each was already possible in Bubble, but only through steps Emmanuel describes as cumbersome.
His advice on both is the same: Look for places in your app where these can replace longer logic, because shorter expressions and smaller workflows are easier to read and easier to maintain as your app grows.
Formatting boolean values in a single expression (15:30)
The third addition is formatted as for boolean values, which gives you a value if yes and a value if no, with access to the full range of expressions including option sets. Option sets in particular had been a sticking point.
It's a small addition, but it turns what used to be one long expression, with a lot of logic packed inside, into something much shorter.
Sign in with Apple arrives for native mobile apps (~16:15)
Native mobile apps built on Bubble now support sign in with Apple. It's enabled under settings and device permissions, though before you start, complete the full setup to comply with Apple's requirements.
End-users get the one-button flow they expect, including anonymous email addresses if they choose. Supporting it also helps keep apps compliant. Apple's guidelines have tightened, and for the past couple of years apps offering third-party authentication without also offering Apple carried a risk of rejection.
Emmanuel also used the moment to restate what native means on Bubble. Apps are built with React Native rather than wrapped web apps, which is what makes features like this available in the first place.
Questions from the community
How to get feedback to the right person (17:54)
A pre-submitted question pointed out that Emmanuel often tells people to reach out to the team, without saying who to reach out to. His answer is to use the support bot on Bubble’s main website. With different teams now owning different parts of the platform, he finds it tricky to point people to one name, but anything submitted through the bot gets routed to whoever owns that area. He also noted that Bubble's investment in AI on the support side has made that routing much faster.
Ambassadors and members of Bubble's Slack community often have direct relationships with a PM or designer, and reaching out that way works well. Those connections aren't a requirement, though, since the general support channel gets feedback to the same place. "Ultimately we cannot fix what we don't hear about," he says.
What's next for the Agent (20:04)
Asked which upcoming features he's most excited about, Emmanuel picked the Bubble AI Agent (beta), with the scaling work as a close second.
The Agent has expanded well past its original scope. It can now work with API calls, plugins, and option sets. "All the [building tasks] that you can do as a human, the Agent can do that," he says. The current focus has shifted from adding capabilities to improving quality and accuracy, with more due in September.
He framed accuracy as the harder problem that the team is now tackling. Agents tend to be enthusiastic about helping, but enthusiasm isn't the same as doing what was asked. He also returned to why he thinks this combination is the right one. AI is good at generating code, but code leaves you with something to read and maintain. AI generating something visual produces logic that's easier to read and harder to misinterpret.
Where visual editing fits as AI improves (22:04)
A related question asked whether visual editing will shrink or disappear as more AI features arrive in the editor. Emmanuel doesn't think so, but he expects how people use it to change, and to diverge.
Some builders he talks to still prefer building by hand, because once you know Bubble you can move very fast by dragging and dropping. Others want the conversational experience they get from tools like Claude or ChatGPT, and those builders will likely use the visual editor more for review than for editing. He sees both as fine, and expects the same split across creation tools generally.
The commitment on Bubble's side is unchanged. "We're very much going to keep investing on the visual part because that's what makes us unique."
Frontend, backend, or both (23:24)
A community member asked whether Bubble's upcoming features skew frontend, backend, or split evenly between the two. Emmanuel put it to roughly even, though it varies from team to team.
The scaling team's work leans backend, since it covers apps performing well at volume and several people working in the same app at once. Elsewhere, the team is working on the Agent generating better-looking applications, both at initial generation and in later edits. Emmanuel described that as something that isn't a frontend feature but leads to frontend improvements.
He tied this back to how Bubble was positioned from the start, as a full-stack replacement covering everything from the frontend framework to the database, so the team keeps pushing on both.
What makes a page slow to load (28:41)
Emmanuel was upfront that this is better handled case by case, and encouraged anyone with a slow page to reach out so the team can look at it from the backend, where they have more information than the builder does.
The general patterns are straightforward. A page loads slowly if it's heavy, meaning a lot of elements or images, though today's connection speeds mean you have to add quite a lot before it registers. The more common culprit is data. If displaying the page requires a chain of searches, where one query decides what a second query looks for, that time adds up before anything appears.
His suggestion is to map the critical path, working out what has to happen before the page can display and optimizing from there.
Native app or responsive web app (31:56 and 41:22)
Asked how hard it is to move a finished web app over to mobile, Emmanuel pushed back on the framing. He'd rather people treat a native app as a product extension than a conversion. Much of the design system carries over, and a lot of logic can be reused, particularly if you've built with reusable elements. But mobile users are usually in a different situation with a different screen, and often want a somewhat different product.
A later question asked whether a responsive web app is enough for mobile users. That one depends on the business. Native apps suit consumer products, and there's something appealing about a downloadable app.
Native is also the only option if you need device capabilities like health data, contacts, or background location. But web apps are faster to build, faster to ship, and let you fix things immediately, while app store submissions and update approvals cost time. If you can start with web, he recommends starting with web.
Bubble compared to vibe coding tools for non-coders (37:36)
A builder asked how Bubble compares to a tool like Lovable for someone who doesn't code. At first the two are similar, Emmanuel says. Both start with a prompt and produce a first draft in minutes.
The difference comes next. On a vibe coding tool you either edit the code or talk to the agent, and for a non-coder that means the agent, every time. Bubble adds a third option. "You can use your mouse and do things in our editor by pointing and clicking."
Once a first version exists, he encourages people to look at how it was built, since workflows and dynamic expressions are far more approachable than code. Fixing a logic issue through a dropdown beats explaining the change to a bot. He was candid, too, that Bubble's Agent still needs to catch up with those tools conversationally.
Shorter answers
The new data tab UI that Emmanuel demoed in June is on hold, behind the scaling work and the Agent. He didn't share a timeline, but said it's still worth doing (33:34).
Sign in with Microsoft is already possible through plugins, and Emmanuel thinks there may be room for a more native option. He'll flag it to the team so it's on their radar, though he stopped short of promising anything (~34:00).
The option sets tab has already improved and there are ideas for more, but it isn't a current priority. Specific pain points are worth sending in, since the design team sometimes runs ahead of engineering on editor work (34:22).
A native request app review action calling Apple's and Google's review APIs would let apps ask for a rating at the right moment. Emmanuel wasn't sure where it sits on the roadmap and will pass it to the team (35:07).
Workflows running in parallel can produce inconsistent data when two of them touch the same record in the same millisecond, which takes a lot of traffic to happen at all. The team is doing groundwork on a fix now, though Emmanuel didn't share a timeline (40:14).
A Bubble app for building on the go isn't close. Emmanuel likes the idea, but thinks the editor suits a laptop or a tablet. Chatting with a bot would translate well to mobile, so the case grows as the Agent improves, and that's where the team is focused (36:40).
The Enterprise Hub isn't built on Bubble, though the success team built it themselves using off-the-shelf pieces to move quickly. That could change. Bubble's success and enterprise teams do build complex internal tools on the platform (30:40).
A Bubble expert badge on Contra is something Emmanuel will raise with Contra directly, though he wouldn't say when, since a few things need to happen first (39:45).
The Bubble Free-For-All challenge winners
Bubble ran the Free-For-All challenge with Contra (43:47), and this round had no fixed category and no required starting point, so any industry and any problem were fair game. Participants had about a week to take an idea to a working app, either from a blank slate or an AI-generated starting point, competing for a $10,000 prize pool across five awards.
Emmanuel called it the most impressive round to date. Most submissions, if not all, were built by one person, and no two apps solved the same problem.
- First place went to Alexandru Trofaniuc for Pushback, a negotiation game where you play a freelancer up against an AI client with a hidden agenda, scored on how much of your rate survives. Alexandru lost $600 the first time he played his own game, then came back and won.
- Second place went to Lauren Shupp for Shared Envelope, a group card and gifting platform where you design a card, invite people to sign, and split the cost of the gift automatically.
- Third place went to Rachel Kavitha for App Aid, a production incident simulator built around an emergency room metaphor, where you diagnose and resolve an incident without touching a live system.
- Best native app went to Akbar Putra Pai for Monian, a math app for children with a companion character alongside them as they work through questions.
- The community favorite award went to Yuri Oleskiv for Deepen, a mobile card game. That award goes to whoever engaged most with peers during the challenge.
For anyone the challenge left inspired to build something, Emmanuel pointed to Launch Lab. It runs monthly, gives you 30 days to take an idea to launch, and includes weekly hours with Bubble's team plus help on marketing and sometimes PR. "Ultimately your success as a builder is our success as a platform," he says.
Emmanuel closed the AMA where he started. Fourteen years in, he sees the scaling and enterprise work as what sets Bubble apart, since people have built and grown actual businesses on it. He also mentioned he's easier to reach than people assume, on social or by email.
Full question list with timestamps
0:02 — Welcome and agenda
2:26 — Read replicas for dedicated plans
4:44 — Enterprise Hub demo
6:47 — Merge improvements (live demo at 8:05)
10:27 — Editor performance on large applications
~11:45 — Deployment speed improvements
12:39 — Transition from enterprise updates to general product updates
13:17 — New :toggle operator and convert to date expression
15:30 — Formatted as for boolean values
~16:15 — Sign in with Apple for native mobile apps
17:54 — Emmanuel often says to reach out to the team, but who should people contact?
20:04 — What are the most exciting upcoming features on the roadmap for developers?
22:04 — With more AI features in the editor, will visual editing shrink or disappear?
23:24 — Are future Bubble features more frontend, backend, or truly 50/50?
~24:45 — Happy 14th birthday: what's changed most, and is there anything you miss from the old editor?
28:41 — What common mistakes increase page load time, and how can we fix them?
30:40 — Is the Enterprise Hub built on Bubble?
31:56 — Once a web app is finished, how hard is it to convert it to a mobile app?
33:16 — Comment: loving the new features, especially the enterprise ones
33:34 — When will the new data tab UI be available?
~34:00 — Any thoughts on sign in with Microsoft?
34:22 — Any update to the option sets tab?
35:07 — Will there be a native request app review action for Apple and Google?
35:34 — Will critical alerts be supported in the native mobile app builder?
36:40 — Is a Bubble app coming so people can build remotely or on the go?
37:36 — For a non-coder, how does building on Bubble compare to a tool like Lovable?
39:45 — When are we getting a Bubble expert badge on Contra?
40:14 — Workflows running in parallel can cause data inconsistencies. Is a native solution for sequential processing planned?
41:22 — Should I rely on a responsive web app for mobile users instead of building native?
43:47 — Bubble Free-For-All challenge format and submissions
46:19 — Winners announced
48:41 — Community favorite award and the Launch Lab invitation
50:54 — Closing remarks
Build for as long as you want on the Free plan. Only upgrade when you're ready to launch.
Join Bubble
