I signed up, and sent you DM. Iām waiting for you to approve me. Right now Iām in view-only mode so I can only chat, but Bubbleās built-in AI isnāt even in the same league.
That, yes, and fact they didnāt need AI to do it, so itās not as if AI is now the reason for somebody to build a focused alternative that Includes all things many of us have been wanting for years.
Yes, AI could potentially speed up the process a bit, but never was a 5 year platform play, as before AI, it was done in 18 months. There is a lot more to building a bubble alternative. But, it is true, Bubble, like most any software company that just leverages APIs and services of other providers (AWS, hosting service etc) do not have very large Moats, and so likely never get to enjoy the concept of keeping competitors at bay and being the only game in town.
Ya I canāt approve everyone right now as I want to ensure itās in an exceptional state for a wider release (nobodyās gonna use it if your first try doesnāt do what you expect!)
But when you have access, Allow edits will become selectable 
It actually is exactly the reason. Consider Kross which lets you convert a Bubble app to code, has all the table stakes features (native AI streaming, listeners for backend events) many top priority on the QOL wishlist (smart table as a component, editor preferences, etc) and of course able to export the code whenever youāre in the mood to do so..
Is Kross building Bubble in code or is it translating a Bubble web app to code?
more Buildprint reviews coming in
we are so back
The decisions it made, either by its own logic or by reading the decisions Iād previously made, were exactly in line with how I build - right down to the naming convention of variable elements. Everything just kind of clicked into place.
5 Likes
Both. For those who appreciate the visual drag and drop of Bubble, the biggest thing keeping them on Bubble is the fact that their app is JSON and not code.
Buildprint is amazing. Well done @georgecollier
What AI provider is everyone using? Iāve been playing with Codex, but heard Claude is really good
1 Like
Most wouldnāt know what to do with code if they had it. And, most business owners couldnāt care less about having the codeā¦they just want an app that works.
I think itās important to realize that sites like Bolt, Lovable, Base44, etc. will go away. The big AI companies will just incorporate what they can do into it. Weāre in the gold rush of AI right nowā¦it will eventually even out to just the big players.
Myself, I would be hesitant to rely on the current AI builders that use the big AIs to build somethingā¦their days I believe are numbered
5 Likes
Canāt wait to get subscribed to this service!!
Only thing is..where are the terms and conditions of use? All data we recieve from clients are under strict confidentiality requirements. This means that we can only use suppliers in our ecosystem that treat the information as confidential information as well (under enfocerable legal terms) and have basic non-licensing terms applyied to the data (i.e. the reciver cannot use it for anything else then to render the service and not form training review sub-lincensing etc.).
Alternatively, hopefully Bubble sees the power of this and purchases it
Thanks for sharing!
everything as youād expect, main thing to keep in mind is that logs are stored in the US so thatās something to consider for EU sensitive applications - but the logs feature is optional
This looks very promising, excited to see the Edit features released.
Does Buildprint cost Work Units, especially the log importing feature?
Could we get a toggle that says āDo not use my data to improve the productā
I understand this is being put out for free and to some extent this helps you subsidize the cost, but I think leaving it on by default and at least allowing a toggle would be nice. If not for free users then at least for a future paid plan.
Maybe in a future paid plan
Thereās no ātrainingā going on at all. But realistically I do need to know how many times users are using the service, tracking errors, seeing usage patterns / the types of things people are doing etc.
2 Likes
Yeah I know, not worried about that really. For me itās just a matter of my business logic, data, etc. needing absolute confidentiality/privacy. I think most people donāt care and wouldnāt enable the setting but for me itās a must.
I mean for data you can restrict the collaborator permissions and for business logic, itās all public anyway except backend workflows
Thatās where most of my business logic is.
1 Like
Better multi-agent orchestration, branch permissions, plan mode
This is a big update for Buildprint!
Since the launch of the editing beta, those with access have found it powerful enough that they now work on multiple projects in parallel. This requires a compatible agent UI that lets you easily monitor your agents in multiple projects.
In addition, a previous limitation has been that you cannot change the view/edit permissions during a conversation.
As a result, Iām releasing the following:
Overhauled project management UI
You can now monitor multiple agentsā projects easily. Itās recommended that you archive conversations when youāre done with them, so that youāre only looking at conversations that youāre actively working on.
Buildprint will make a notification bubble sound (see what I did there) when an agent is done, and it will show as unread.
Better permissions for agents
You can now update the view only / allow edits permissions during a conversation. This allows you to begin a conversation with a plan, and then grant it permission to edit when you are happy.
In addition, the branch permissions selector lets you optionally constrain the agent to certain branches. It will automatically be granted access to any branches it creates itself.
Better branch visibility
You can now see which branch an agent is working in at the moment in the header of a conversation. This lets you easily check its work on the correct branch and verify itās working where you expect.
View plans and attachments in the UI
Faster edits
Previously, large apps could take a long time to process each edit which made Buildprint slower. We have implemented some optimisations here that make it much, much faster for these apps (about 95% faster).
Top up Buildprint credits
For users without connected ChatGPT/Claude subscriptions, weāve added a simple $20 top-up functionality that lets you add AI credits to your Buildprint account. We charge credits at cost so it just lets you use AI at the model providerās price if youād prefer not to maintain a ChatGPT/Claude subscription.
Logs updates
To be friendly to Bubbleās platform, weāve updated our logs processing. High traffic apps will see logs synced very quickly, although low traffic apps may have log sync delayed for up to 12 hours. This may change in future, but Buildprint accounts for a significant portion of Bubbleās total log fetch volume so we must respect their own platform limits. We have some hard limits we need to respect because of their provider - in the future, logs will become a paid feature (free tier + usage based) because, well, it costs a lot of money to ingest 2.5 billion logs per month!
Notes on wider beta release
The beta release has been going very well - I am intending to keep it fairly narrow in terms of users for now, as I must ensure I have the capacity to respond to feedback. You can follow along with the beta testers experience on the forum and social media.
Usage is increasing significantly even in this very closed beta:
We experienced a bit of downtime yesterday due to a backend provider outage, and I was getting spammed over it, which means people are using Buildprint!
7 Likes