@_adishj@motion_so That looks really good. We tried @Remotion but it needs lots of tweaking to even make it look good enough. We would like try @motion_so.
The AI industry has become so LLM-brained that people forget classical ML still beats the SOTA multimodal language models at plenty of real-world tasks.
We use a classical ML model for intent detection.
@programmerByDay We are building HeyZinc, a tool that lets founders talk directly to visitors on their website. Helps a lot to get first few customers.
@daicandev Cause we let you call visitors on your website.
User taking too much time to complete onboarding? Just call them and ask why.
No need to book meetings or anything.
@garrytan Its funny cause we also realized something similar and built the two loop architecture but for the AI Voice agents. Straight from our slides: