> A simple example is substituting long strings the LLM often refers to. For example, when a user uploads an attachment, we give v0 a blob storage URL. That URL can be very long (hundreds of characters), which can cost 10s of tokens and impact performance.
> Before we invoke the LLM, we replace the long URLs with shorter versions that get transformed into the proper URL after the LLM finishes its response.
So, instead of fixing your blob URLs, you effectively run a regex to search and replace on the prompt before you feed it to the LLM? That does not seem like high quality engineering.
> In addition to text injection, we worked with the AI SDK team to provide examples in the v0 agent’s read-only filesystem. These are hand-curated directories with code samples designed for LLM consumption. When v0 decides to use the SDK, it can search these directories for relevant patterns such as image generation, routing, or integrating web search tools.
Curated or created? If you "curated" them what is the copyright license on those examples? Are they just copied into the project?
Remember when programming was about creating useful libraries of code? Now it's about sequestering them inside of an LLM and then charging insane amounts of money to convince a machine to sometimes copy them into your code for you.
Just.. wow... what are we even doing? Prepare for a very fragile future.
Whatever their secret sauce is (model, system prompt, or harness), v0 creates some of the most beautiful designs and mockups when I ask it to. And it looks really polished.
Even better than anything I’ve seen from Claude Code.
So I use it a lot for inspiration for screens and even have used it for proposals.
I feel like using sequential transformers generating code feels like a brute-force solution for UI generation.
The "obvious" path forward for frontend assistants is to move away from raw code generation toward some domain-specific language. UI is inherently structural - it should be expressed through component hierarchies that implement pre-defined design guidelines (with colors, margins, border radius etc being defined outside of the core model, and applied as a kind of theme to the model's output).
If we define the UI as a composition of VStacks/HStacks and predefined components, we can use diffusion models to generate the layout, and consistently apply the 'theme' afterwards. It's a much cleaner abstraction than asking an LLM to hallucinate valid CSS classes.
5 comments
[ 2.8 ms ] story [ 26.8 ms ] thread> Before we invoke the LLM, we replace the long URLs with shorter versions that get transformed into the proper URL after the LLM finishes its response.
So, instead of fixing your blob URLs, you effectively run a regex to search and replace on the prompt before you feed it to the LLM? That does not seem like high quality engineering.
> In addition to text injection, we worked with the AI SDK team to provide examples in the v0 agent’s read-only filesystem. These are hand-curated directories with code samples designed for LLM consumption. When v0 decides to use the SDK, it can search these directories for relevant patterns such as image generation, routing, or integrating web search tools.
Curated or created? If you "curated" them what is the copyright license on those examples? Are they just copied into the project?
Remember when programming was about creating useful libraries of code? Now it's about sequestering them inside of an LLM and then charging insane amounts of money to convince a machine to sometimes copy them into your code for you.
Just.. wow... what are we even doing? Prepare for a very fragile future.
Even better than anything I’ve seen from Claude Code.
So I use it a lot for inspiration for screens and even have used it for proposals.
The "obvious" path forward for frontend assistants is to move away from raw code generation toward some domain-specific language. UI is inherently structural - it should be expressed through component hierarchies that implement pre-defined design guidelines (with colors, margins, border radius etc being defined outside of the core model, and applied as a kind of theme to the model's output).
If we define the UI as a composition of VStacks/HStacks and predefined components, we can use diffusion models to generate the layout, and consistently apply the 'theme' afterwards. It's a much cleaner abstraction than asking an LLM to hallucinate valid CSS classes.