Monthly Community Update -- May 2024

Hi all,

This is our May community update. You can read last month’s update here.

April was a busy one, as the entire Bubble team gathered in NYC for a whole week mid-month. We do this twice a year to remind ourselves we’re working with humans, not just squares on video screens. A lot of the work we do is remote, and it’s important to reconnect with each other as people to build relationships that can withstand the inevitable frustrations that come up working on an intensely demanding product. In addition to connecting with each other, I also think it’s essential that we connect with you all: Last October’s BubbleCon was a really powerful experience, and one of our big priorities this year is creating more opportunities for connection — via the forum and social media, via video, and in person. A lot of our team read the forum regularly, and I’d love for you to hear directly from them!

Before I get into this month’s updates, a quick note: The Bubble Developer Certification exam is going to be undergoing planned downtime for maintenance the week of May 13. If you’re planning on taking your exam then, you’ll need to take it the week before or the week after instead. (If your exam registration expires that week, you’ll get an extension.) More exciting updates on certification coming soon!

On our minds: Uptime and reliability

As many of you are aware, we had an hour of downtime last week that affected all main cluster apps, plus some impact on Dedicated functionality that relies on the main cluster.

@payam.azadi , Bubble’s director of platform engineering, published an official postmortem on the incident on Thursday. You can read it here. The postmortem details not only what went wrong, but also our learnings going forward. We also had two smaller incidents, one Monday, one today, that we are currently working on diagnosing and remediating.

In addition to what Payam wrote, I’d like to add that we’ve had a lot of internal conversations about reliability, both leading up to and as a response to Thursday’s outage. We know how frustrating bugs and downtime are, especially given the fact that many of you make your livelihoods via Bubble. I also know “we’re working on it” is not a particularly satisfactory response, even if it’s a true one. I want to share some context on why this isn’t something we can change on a dime, as well as the progress I’ve seen and why I’m optimistic here.

There are two main flavors of incidents that we respond to: scaling issues, caused by user and data growth pushing the limits of our systems; and regressions, problems introduced when we ship new code. With the former, we’ve come an incredibly long way from where we started: Our original infrastructure was built by Emmanuel and me, and it would collapse under the weight of simple things like a user running a recursive workflow. Our current infrastructure is orders of magnitude more sophisticated — but as fast as we improve it, we have users building more and more complicated applications and finding more extreme edge cases. We have been carefully putting in limits, but we also can’t blindly restrict what users do with our systems without breaking existing apps. That means we have to take it a step at a time while simultaneously replacing our least-scalable systems with new ones that can handle orders of magnitude more load, as well as allowing user apps to scale further and faster.

The easiest way to prevent regressions would be to not ship new code, but given our scaling challenges, we don’t believe we can afford to stand still. We also know it is critical to continue improving the product, because if we stagnate as a business, we won’t be able to afford to invest in our infrastructure, and it will kill the platform as surely as extended outages would. So, our strategy on regressions is to continue to build automated tests to close off all known ways we can break things, invest in our continuous integration to be able to deploy code more quickly and safely, and improve our alerting and monitoring to respond to problems faster and catch them before they have user impact. Finally, we are iterating on our team process and culture to strike a balance between making bold improvements and doing it in a way that’s appropriately cautious.

I’m optimistic because of the changes we’ve made over the last six months:

  • Our time to respond to incidents is much faster, as is our average resolution time

  • We now publish every main cluster incident that involves an emergency response to our status page. This is a mixed blessing, because it means that the whole user base sees incidents that may have only impacted 10 apps. But we would rather be transparent, since we know how frustrating it is to discover your app is down but not see any acknowledgement that we are aware of the problem.

  • In addition to Payam, we’ve hired three experienced engineering managers (two of whom started in April) who are bringing in technical and process best practices

  • We’ve increased the scalability of our bulk actions, especially “Schedule a Workflow on a List”, which reduces reliance on potentially-fragile recursive workflows

  • We’re undertaking some cutting-edge technical work around our cloud environment and the way we manage databases that should have major impacts later this year

I wish I could promise you that we’ll never see another incident, but that would be unrealistic. What I can promise you is a trajectory of continuous improvement. This is extremely important to me personally, to Emmanuel, and to the whole Bubble team, and we are right there with you pushing for Bubble to be a stable, scalable platform.

Changes we made this month

Our updated agency directory is live! You can now sort and search by more filters, including budget, services, location, and language. You can also click through to view every agency’s profile, which features their portfolio, links to their website, and more. We hope this helps as you look for the right expert to help you with your next project — the talent at these Bubble agencies is top-notch.

We also made improvements editor load times by caching the editor plugin list. This change improved the 80th percentile editor load time by 15 seconds.

Amid the strong positive feedback on last month’s update to the elements tree, we also heard feedback that the team is working to address. To start, it’s now easier to open the property editor because you only need to click once on an element in the tree to open the property editor.

In case you missed it, the new Getting Started With Bubble YouTube series hosted by Gregory John is in full swing. There are nearly 38,000 views on the playlist so far! We add a new section every week — subscribe to the channel so you don’t miss the last two drops this Monday and next Monday.

And last but not least, we launched the official What’s Bubbling newsletter for founders! You can subscribe here or via your account’s email preference settings if you didn’t get it. (If you’re a freelance or agency developer, you should already be subscribed to the Bubble Developer’s edition — if not, the link to subscribe is the same.)

What we’re working on

  • Reliability: Our platform team is hard at work implementing the action items Payam laid out in his postmortem on last week’s outage.

  • Mobile: This month, the team launched an internal alpha of our native mobile editor! To prep, they focused on wrapping up core user-facing functionality like native mobile gestures and native mobile components. Now they’re getting ready for a private alpha by reviewing early feedback and fixing bugs, plus breaking ground on new work like native alerts and support for multiple icon libraries. To make sure you’re on the waitlist for the public beta later this year, sign up here.

  • Version control changelog: This feature will be entering a private beta in the next few weeks. Changelog enables you to view all changes made on a branch, who made the change, and when the change was made — making debugging, quality control, collaboration, and understanding merge conflicts and changes even easier.

  • Elements tree updates: As I mentioned above, we’re making some updates to the new elements tree thanks to your feedback. This includes saving a renamed element automatically when you exit, adding show/hide children in the context menu, adding a clear function to the search bar, and creating better drop zones within the tree.

  • AI: The upcoming AI page designer has been undergoing private alpha testing by hundreds of new users and select agencies. Alpha testers have generated 400+ unique pages so far and have provided meaningful feedback. Stay tuned for an open beta in June, plus an upcoming event announcement! We expect this initial feature to be most helpful for new users; agencies have told us that it’s probably not a game-changer for their day-to-day work just yet. Still, this is an exciting step forward because it informs the future of how AI and no-code work together on Bubble.

  • Observability: The team is continuing work here, specifically on ways to track progress, understand outcomes, and troubleshoot issues for workflows scheduled. We’ll have more details to share soon.

  • Plugins page makeover: We’re working on a long-awaited redesign to both /plugins and /plugin to make it easier for you to find the right plugin for your needs. You’ll be able to search with a new sorting functionality and filters, and plugin developers will have new categories to classify their plugins, like PDF, chart, chat, calendar, and image. Note: When this redesigned page ships in late May, any plugins that aren’t built in the new responsive format will be delisted.

A peek further down the roadmap

The AI team is excited about the way our AI page designer (mentioned above) makes it easier to design frontends on Bubble. Once that feature is officially in your hands, they’ll be turning their attention to the next exciting next step: AI-assisted dynamic backends. In other words, the goal is for the text-to-build AI feature to not only build you a static page, but also fully flesh it out with a matching database and workflows.

We’re also scoping the future of the workflow tab, which is overdue for a UX overhaul. This will be a big step toward our broader initiative of modernizing the editor to make it a first-rate, highly usable experience.

Finally, we’re making ambitious changes to improve database performance by changing the way we store user data behind the scenes, with the goal of making it blazingly fast and scalable. More to come soon!

Applications for our next Immerse cohort are now open!

Immerse is Bubble’s 8-week, fully-funded pre-accelerator for founders. Exciting news: This cohort (our seventh) is called Immerse Impact and will support people who have social impact at the heart of their missions. You can read more about that focus here and apply here!

We’re also looking for more volunteer product mentors to coach this cohort as they learn to build on Bubble. If you’re interested, you can learn more and apply here.

From the Bubble blog

Here are a few stories worth checking out, if you missed them:

Plus, we just published a Showcase story about Farie, an auto marketplace startup that recently raised a $4.5M Series A from one of the largest insurance companies in Switzerland.

Team updates

We had six new hires start this month! Welcome to the following folks:

  • Federico, our new community operations and moderation manager

  • Shen, joining us as an enterprise account executive

  • Kyle, joining us as a sales development representative

  • Jeffrey, our new lead data engineer

  • Joe, joining us as an engineering manager

  • Gongjing, also joining us as an engineering manager

We have plenty of open roles — check them out here.

That’s all for this month. See you in June!

— Josh and Emmanuel

36 Likes

Very good feature :grin:

No surprises there, a quick scan of the forum would tell you that there are no power users at all that are happy about Bubble x AI. AI backends/workflows sound good in principle but what’ll actually happen is we’ll spend more time debugging them or trying to work around them than we would just building by scratch. You’ll get a bunch of people here telling you to move this workforce into the editor modernisation team :upside_down_face:

Like this :+1:

Please look at WeWeb for this! They do it very well - though, their design would imply branched workflows which would be killer but understandably not on the short term roadmap.

Reliability - SLA anywhere? I have no doubt ‘changing the way we store data behind the scenes’ will involve a bunch of accidental breakages and downtime caused by implementing that. We’d rather it works and works slowly than works faster and intermittently.

10 Likes

I think the reliability of the platform should be the first focus and is more important than shipping AI and other fancy things. Personally, I am very frustrated as these sorts of outages are a new normal.

14 Likes

Please, focus on this. We don’t need all these whistles and bells like AI and nice elements tree if there are problems with our apps being down or slow.

12 Likes

I think we all agree.

Prefacing this with the fact that I’ve never run a traditional code SaaS business, reading between the lines, it seems that they’ve probably of put everyone that has the knowledge to help with reliability on that task already and can’t do any more to accelerate progress on that front. For example, it seems difficult to put recent hires on improving platform reliability as they lack the institutional knowledge of the senior developers that have been there for ages and can actually contribute meaningful fixes.

Still, reliability aside, it would be nice to see the AI efforts directed to other areas of the product…

1 Like

THANK YOU! That tiny thing was a huge nuisance.

Also thank you for the transparency on the outages and scaling issues.

3 Likes

Hey @josh, is a way for developers to extend the native capabilities planned - i.e. something akin to plugins for Bubble web apps?

4 Likes

Any chances of some new features related to WU consumption, so we could start to build optimally within WU consumption restrictions, like electing which data fields to return when performing a search?

Also, any plans to get the engineering team in for a full run down of ALL things we are charged WUs for so that a complete list of costs could be published?

Any chance there will be some changes to the way URL paths are currently not decoded, nor can we use the get data from URL dynamic expression and elect a path or path list item and declare it as a ‘thing’ and for it to be recognized if the value is the things slug or unique ID, just like the way URL parameters currently work?

Nice, I can build a bit more quickly. Any improvements being made in terms of page load speed and performance so that we might be able to get a blank page with a 100% score for Performance on Google Page Insights? I’d love to see optimally built pages get 90+ scores on Performance by Google Page Insights.

I would imagine performance, reliability and mobile would be more imperative for the success of the business. If an AI can build me an app that goes down often and doesn’t have the page load speeds necessary to keep users from jumping off the app, there will likely be no benefit to the fact that AI sped up the process of building the app.

Having better page load speeds will help us a business owners (not agencies but app owners) by saving us money because, if users are waiting for 3 seconds for a page to load, and 36% of them leave before it loads, well, then I’m throwing $36 in the trash for every $100 I spend on advertising via Facebook and Google. It also might even help, not just in saving money wasted on advertising, it could increase our revenue as having that extra 36% interested potential user base maybe 1-2% of them are converted into paying users.

13 Likes

I friend of mine is building a similar product like bubble (LowCode). He recently told me that bubble code base is messed up and I didn’t agreed until now. Yes, I believe bubble might be putting all the efforts to resolve these outages but I guess it would be better if bubble stop trying to do all the stuff by themselves. We should be able to host applications on our servers and yes we will be glad to pay a sub fee.

Thank you for this monthly update. always appreciated

I would like to say:

What we’re not working on

I will resume this: everything thing asked around WU
like: Select which fields data to return
Improve logs and chart to make it easier to investigate WU
WU / users
Slugs improvment
Issue with 0 in expression like :count is 0
Average WU / workflows
And probably a lot more than that suggested here and on ideaboard.

The only and first WU related thing released is the new Schedule on a list. We should have seen a lot more improvment on this side but I don’t see any focus on that from your team. But it should be on top of all other thing to work on.

Thanks

12 Likes

There’s a lot of low code solutions. Most of them are, in the end, far from what Bubble NO code can do.

4 Likes

:yawning_face::yawning_face::yawning_face:

:fire::fire:

:fire::fire::fire::fire: More of this plz

Hi @sudsy, great question! Can you please email us at community@bubble.io with your use case so we can share with the team? Thank you

1 Like

Are you guys ever going to fix the table element? This was released 10 months ago and its still not useable. How many people asked for this feature vs how many people have asked for an update to the plugin page?

9 Likes

some good work here nice

Great work!

Thanks for this update.

Excited about the database performance improvement, and uptime/reliability!

If I may also share an item in my wish list, I would love to be able to know WU consumption per user

3 Likes

Would love to understand the markers Bubble use to prioritise their roadmap.

Seems targeted towards investors and ‘new bubblers’.

Why are things like native wrappers getting progress when there are perfectly good third party options available? I guarantee there’s an extra cost associated with this, which would prove me right about the roadmap being targeted towards investors. Have the unintended consequences of losing plugin builders been thought of?

Also this has me in stitches: “ The easiest way to prevent regressions would be to not ship new code” - wouldn’t the best way be to do proper regression testing?

2 Likes

some great updates here.

Perhaps AI could be used in a different and helpful way. For example, as an analysis tool with advice on better structuring. An example of advice that would be helpful: Here is a suggestion on how you could save WU: Explanation: xyz.

Or: Your database contains xyz, you could make a query here so that the loading speed of your page is increased. I believe that this area can already bring significant improvements, but used correctly.

9 Likes

Has native mobile moved from “mid 2024” to “late 2024”. Can you give more detail on timeline to help people plan?

2 Likes