GANT and Chat Components are 4 month overdue by now while all you do is trying to get an AI working nobody asked for. Now you stopped communicating. When you continue this path you will lose more and more users. The AI people will never choose Weweb over lovable and the others that are purposefully build for that. AND limited by that! You are not an AI First platform and by desperately trying to become one you will fail. Iād really like you to commit to your strengths and that was an easy to use builder with a lot of granular control and smart and simplified ways to talk to your backend.
I agree regarding the overdue, but Iād like to highlight that AI is a very critical feature as it significantly saves time when building forms and handling responsive customizations. This is especially true when working with REST APIs, where a lot of mapping is usually required. The latest AI control context is much improved for now.
The Chat component is also very important, and I hope it will be completed soon.
For Gantt charts, I suggest looking at quickchart.io, which helps address gaps that many low-code builders have.
Have also a look on this thread
At first, I was skeptical about the AI, but when the major bug occurred, its value became more apparent by contrast. I believe they now have a better solution in place, and hopefully it stabilizes soon.
I have to agree with the above. What AI could do at the very beginning was quite enough for all the tasks for which it was really useful: editing forked components and creating new components.
In my opinion, it makes no sense to make an analog of Lovable or Replit, because we all love Weweb globally for the 3 most important things.:
- for no-code experience;
- for outstanding flexibility;
- for full control over the project.
Unfortunately, AI completely negates control, because it changes properties that can no longer be accessed through the editor interface. This means that you need to research the code. And if you need to explore the code, then all the charm of Weweb is lost.
And Iām not even talking about the fact that sometimes it breaks something very unobvious, which then takes a lot of time to debug.
P.S. And we all really need overdue elements.
Hey everyone,
I see the potential of AI in WeWeb, but right now it creates more problems than it solves. A few examples from my side:
-
Loss of predictability: When I ask AI to generate or adjust something, I often donāt know what it will actually change. That makes projects harder to maintain in the long run.
-
Hidden changes: Sometimes the AI modifies properties deep in the code that are invisible in the editor. Later I have to debug issues that I didnāt create myself.
-
Workflow disruption: Instead of speeding things up, I end up spending more time double-checking and undoing AI actions.
-
Priority mismatch: As others said, many core features (Gantt, Chat, better data components, etc.) have been overdue for months. For a builder like WeWeb, these are much more important than chasing after āAI-firstā positioning.
-
Learning curve: New users get confused because the AI introduces complexity that doesnāt match the promise of a simple and intuitive no-code experience.
On top of that: I tried today to build several components with WeWeb AI. Not a single time was the AI able to complete the component without throwing a runtime error. I experimented with different approaches, but without success.
There have been no real updates from the team ā no clear communication about the current mood, roadmap, or plans. This creates an uneasy feeling about where WeWeb is heading. New users who might be excited about AI will leave as quickly as they came when they face these issues.
From my perspective, itās crucial not to overlook the developers who use WeWeb as a serious, professional web app tool. These users donāt need flashy features that break; they need stability, reliable components, and trust that their projects will not suddenly be disrupted by hidden AI changes. If WeWeb keeps focusing on control, stability, and overdue core features, it can continue to be unique in the market. If not, it risks losing its strongest user base.
Iāve created some tickets, but itās hard to wait for the investigation. A basic example: to avoid creating tons of unexpected vars/wfs I created a new project and asked to make a chat box.
And hereās an example of programmatical binding:
Iāve already fixed all the stuff manually. But seriously, I can no longer rely on your AI.
- It fails to create a component for no reason.
- It fails to follow existing design rules and instructions like āuse only vars I selected for colorsā (vars are properly named).
- It canāt fix any issue with the component now. Instead, it can update just AI.md and do nothing to the code. Or recreate the layout completely when you ask it to fix something.
- It reverts your changes (like after it created the quotation in a chat and Iāve fixed the layout manually, it restores its original bugged version when I ask for some modification).
- Sometimes it just says āInternal Server Errorā and you have nothing to do with it.
- It creates hidden settings which are not visible in the editor UI (like it created a border which wasnāt shown in UI and the only place I was able to find it was ācreate a styleā where all css props are shown).
This experience makes AI unusable and unpredictable. For 20$/month I can get unlimited ChatGPT or Cursor. Here we have this limitation for tokens and the result is reeeally bad.
I appreciate your tries to do our experience better, but sometimes it looks like thereās no testers in your team.
Iām also afraid the latest AI update wasnāt properly tested before being pushed. Iāve actually really enjoyed the AI when it comes to creating formulas, both weweb formulas and js formulas, and the previous version was a big step up from before. However the latest version couldnāt even do ācheck if user is logged inā, it literally did something like ātoBool(,)ā. Not that I donāt know how to make that one myself, but if it canāt even do that, then something is definitely broken!
The new import to Figma plugin looks really good though, I think itās something extremely useful for many people that will be starting out with a new project and are used to having a design>develop workflow. I just imported a small element and that went flawless!
All we need now is SSR, haha - a lot of the complaining would stop 
Feedback for updated elements:
Rich Text Input
- We need global settings for font and colors. Thatās nice to have override options for selected tags, but usually thatās unnecessary.
- Some translations are not accurate.
- Somewhere thereās no word ābackgroundā for color (Quotes, Code tags).
- āPair/Inpairā for table rows. Odd/Even, Main/Alternative, Primary/Secondary. No need to avoid commonly used slang.
- Still no default āFocusedā state. We have to create a component, make a āFocusedā variable and set styles based on this. Please handle it on your side and give us similar functionality we have for inputs.
- Menu color is not bindable. Custom menu doesnāt show all available buttons.
Uploader
- Again, too many settings. Actually we need 2 global settings: bg color and text color, as usual.
- No information when we delete item from the list of uploaded files. we need similar onDelete action or modify the event in onChange to show old and new states to be able to handle it. Now I have to hide this uploaded files section (why canāt we disable the file list area?) just after the upload by cleaning it to create an UI for listing and proper deletion from the database.
to be updated. I hope youāll react and fix it.
On this same thread, a generic variable mutation listener event workflow would be really cool
when variable X changes, do A=>B=>C