Vibe-coded this metal shader distorting a SwiftUI layer in real-time. It uses SDF fields to define card geometry, merges a touch blob via smooth-minimum for the gooey effect, then displaces pixels toward the tap with spring physics and ripple.
What do you think?
Truckers shouldn't fill out 8 form fields to find a load.
I am working on an app where they type "flatbed out of Houston this week" and an LLM maps it to structured filters via tool-calling and get them streamed results.
Guys, I ended up using a hybrid parser:
local regex parser first, then CommandCode LLM function/tool calling as fallback.
The LLM does not query the DB directly.
It produces structured filters through tool calling, then my backend searches pets with normal TypeORM database queries plus haversine distance filtering.
Need help!
Let’s say I have DB of 10k pets, in different states and different types.
My app users want a feature where instead of manually adding search filters they just want to do a prompt like “Give me brown huskies in Texas not more than 2 years old” and get the results.
I know ChatGPT API key can do it but what are my most cost effective and practical options?
Need help!
Let’s say I have DB of 10k pets, in different states and different types.
My app users want a feature where instead of manually adding search filters they just want to do a prompt like “Give me brown huskies in Texas not more than 2 years old” and get the results.
I know ChatGPT API key can do it but what are my most cost effective and practical options?