Alina Latinina
I curate and critique the Internet of Things.
New to Contra
Alina is ready for their next project!
Unfortunately, I won't be submitting a project, but I would like to share my thoughts on Bubble as I was using it for a day.
I was building a functional app with Bubble AI agent, a GDPR privacy tool - an easy way for humans to ask for their data to be deleted from companies, and a way to track it. The database, generated email formats, the time tracking, all of that.
I went in as a builder, but ended up rage quitting, and here's my emotional journey, in case that helps anyone, and also I would just like to vent :)
I was very excited for Bubble, because out of all the popular vibe coding apps now (lovable, bolt, replit, cursor and more) they're doing something different.
These apps can build out a whole world out of a single sentence and it's all fun and games until you need to fix something in the end, but you're stuck in an endless loop of prompts trying to fix it, but you do not see where the problem is. Trying to read the script, going to Claude or Chat to ask for help, - it gets heated.
Bubble leads with control without sacrificing the speed - every change you can see and refine. No code. You see what's happening in front of you. Comparing it to other apps - transparency and control are in the lead - I liked it.
However, when I was building - the agent said that it all was there - the companies, the data I fed it, dynamic deadlines - but it wasn't on the canvas. The placeholder data was there, and after 3 rounds of prompts trying to fix that, I didn't get the result I was asking for - the trust is lost. The speed is compromised. Again, I crawl back to Claude to ask what is going wrong.
Even a simple change from the AI agent to say "here's what I changed; confirm it landed", instead of "Done!" while it is not, would make me trust it more.
Bubble could own a story that is both more accurate and more differentiated: not "a functional app from one prompt," but "the only AI builder that gives you a foundation you can see, trust, and control". The gap between what is promised and what is actually delivered is a gap that won't survive this market.
Also, on a smaller note, but also important one - we are living through the dread of AI bubble bursting. I went in looking for tutorials, but all I got was tutorials on the economy crashing. Maybe rebranding the name or being more deliberate about it would be a good move, because as a user I wasn't reading your roadmap, I was looking for some damn tutorials.
I'm attaching the screenshot of the project I was working on - maybe I am stupid and didn't see some simple button that fixes this - but also thought it's worth to share the experience. thanks for reading all that.
1
159
A room of my work - Panopticore - a Substack focused on writing about modern internet, digital privacy and culture.
0
44
I created this brand in a week after my Reddit post went viral.
1
65
My personal website, made into a coherent collection of random works.
0
51
My talk on Digital Privacy + Culture in Cannes
1
67
My article on Digital Privacy in modern age on Rage Magazine
3
94
A chatty podcast about everything humanity and tech. Anti podcast branding.
https://www.youtube.com/watch?v=Cw1Yo3JFDx4&t=1712s
1
68
People are tired of being optimized.
There is a real shift towards softness and simplicity. Finding joy in technology in a way we did in the early 2000s.
BLEEP ME! ♡ is a simple reminder app where you schedule little notes to yourself.
Write a little note, pick when, and BLEEP bleeps you when its time.
It does less on purpose. Less to manage, less to dread, less between you and your life.
Simplicity is the radical choice in an age of overwhelm, and it’s also a product that feels fun to touch.
It’s slow, it’s pink, it’s encouraging.
https://bleepme.lovable.app
https://stitch.withgoogle.com/projects/14252840006577992145
*Workflow*
I used Claude first to articulate the idea I wanted, then I pasted it into Stitch letting it start from the bare minimum. It was a bit of a shitshow, mostly from my side because I didn’t have a clear picture of the design I had in mind. After creating a DESIGN.md (http://DESIGN.md) in Claude Design I uploaded it to Stitch and used 3.1 pro and it went almost seamlessly after. Stitch is definitely better at executing than helping ideate. Then I exported it to Loveable and added a couple of UX corrections.
*Notes on first time using Stitch*
- A little confusing the first time - I wasn't sure where things were going to pop up on screen. Some signposting for new users would help.
- The window sizes keep changing on their own, which is disorienting.
- I'd love to work on a single canvas. Keeping every past version visible by default feels wasteful and actually makes me less likely to iterate. Let me open past versions only when I deliberately choose to.
- If past versions do stay, move them up/out of the way automatically so the current work stays front and center.
- I pressed "Go back" and it re-uploaded everything I'd deleted - still no idea how to go just one small step back.
hope you enjoy! 💕
Linkedin post: https://www.linkedin.com/posts/alinalatinina_i-did-a-little-weekend-challenge-on-stitch-activity-7468003570747850752-0vJA?utm_source=social_share_send&utm_medium=member_desktop_web&rcm=ACoAABTagGoBSWeMQtQaz6TDzJZEKHC4EmArqZw
4
4
226