Context on the use case / what I’m trying to achieve
As a developer who regularly works with code repositories, I want to streamline the creation of frontend interfaces. I’m proposing a tool similar to v0 or Replit that can transform natural language descriptions into functional frontend interfaces for workshop and slate development, allowing developers to describe components, layouts, and interactions in plain English and have them automatically implemented as code.
Why I can’t do it today / work-arounds
My current challenges and workarounds include:
Workshop’s widget-style nature creates complexity that’s difficult to manage efficiently when building interfaces
For slate development, I’m forced to delegate frontend tasks to external AI tools like Claude 3.7 or Gemini 2.5 Pro since the current AIP assist isn’t optimized for frontend development
Context switching between our platform and external tools disrupts workflow and reduces productivity
Manually implementing widget configurations and layouts is time-consuming
The disconnect between natural language descriptions and widget implementation creates unnecessary friction
These limitations force me to use multiple tools and platforms when a single integrated solution would be more efficient.
Why this will make a huge difference for my business
A v0/Replit-style natural language to frontend interface agent would:
Eliminate the need to use off-platform AI tools for frontend development
Significantly reduce the time spent configuring widgets in workshop
Provide specialized frontend assistance that current AIP tools lack
Enable rapid prototyping and iteration directly within our development environment
Keep developers in their workflow without context switching to external tools
Increase productivity by focusing developer attention on logic rather than UI implementation
I am working on a version of what you are asking for in VS Code Workspaces. Specifically working to preconfigure the open source Continue extension to have knowledge of your OSDK and agentic programming for front end apps. (This also works for Python Transforms but you’re not asking about that).
Ensure your organization has Anthropic models available in Control Panel as these are currently the best for agentic programming. Gemini 2.5 pro will be available shortly.
From there we will automatically install the Continue extension, preconfigured to have knowledge of your OSDK application. Available today for Claude Sonnet 3.5, 3.5 v2 and 3.7, there is an agent loop that can autonomously create front end applications for you.
I would love to hear any feedback that you have, please let me know on this thread if you get any issues or have any suggestions.
thank you so much for sharing this feedback! I wanted to let you know that this is exactly something we’re currently experimenting with on the AIP Assist side - so it’s great to hear that this would be of interest to you. Do you have other places where you think this sort of “generative offering” could be useful, or is it mostly code->FE components (Workshop, Slate) that comes to mind?
@mkindblom Is it possible for me to get beta access to this feature? would love to experiment with it and post content if permissible. Furthermore, I do believe there is improvement opportunity in the solution design AIP assist implementation. Currently it’s a one shot recommendation from AIP assist, where at times the recommendation is not entirely accurate due to the need for clarification. Including a conversational feedback loop to iterate on solution designs generated would be extremely helpful and expedite the solution design process. I will continue to play around and brainstorm to identify more areas where generative offerings would be useful and more importantly impactful. Thank you!