Jagath Vytheeswaran

For a long time, Yann LeCun and others believed in gradient-based planning, but it didn’t work very well … until now. Here’s how we did it using incredibly simple techniques. But first, an introduction to gradient-based planning: 🧵1/11

Lunabill (@getlunabill) is the AI call center for healthcare billing teams. They've already partnered with leading US healthcare systems to automate 50,000+ phone calls, saving them over 20,000 hours.

Vogent just launched Vogent Turn 80M, an open source multimodal AI model for turn detection in voice conversations which beats all existing turn detection models right now. Voice AI often struggles to know when you’ve finished speaking versus just pausing, causing awkward interruptions or sluggish responses. This model eliminates this problem entirely. > Multimodal: combines audio signals and text context > Ultra fast: 7ms inference on T4 GPU > Highly accurate: 94.1% accuracy rate > Real time ready: works with batched inputs > Easy integration: simple Python API > Production ready: You can start building today. Available on GitHub and Hugging Face.

.@vogentai just released Vogent-Turn-80M: an open-source, state-of-the-art turn-detection model for voice AI. It improves any voice AI stack by detecting when to wait and when to speak, reducing interruptions and making conversations more humanlike. blog.vogent.ai/posts/voturn-8…

@SukhaniShri 🫡🫡🫡

Happy Diwali!! 🪔 Wishing you AGI, cheaper tokens, Nasdaq ATHs every month, and GPUs for everyone! 🧠💰📈 Let the lights shine bright - on screens and in life💡🔥

JetBrains is no longer behind. @firebender_com just launched the first-ever background coding agents for all JetBrains IDEs. These coding agents are incredibly intelligent, have isolated workspaces, and don’t require any cloud setup.

🗺️ Can LLMs navigate San Francisco purely from memory, without Google Maps? I tested spatial reasoning of the 19 top LLMs. Honestly...they were all pretty bad. Best in cities: o3 (38%) Best on highways: grok-4, o3, o4-mini (62-64%)

made a daily news podcast hosted by joe rogan and trump using ai (perplexity + vogent voicelab)

has anyone actually given money to an npm package because of this message? genuinely curious

@elsewhere_today chai

Everyone in SF complains about how every café closes by 5pm and there’s nowhere to work. Introducing Elsewhere: the world’s first late-night café coworking chain for builders, creatives & digital nomads. Launching July 24. Comment your favorite drink & we'll DM you the invite.

not super reliable (yet) but a better voice design method imo than prompting voice design with text

I discovered this by accident because a customer needed an Indian accent voice ASAP and I couldn't get a voice actor so I did an impression of an indian person and cloned it, and it ended up creating a pretty authentic Indian voice (that didn't sound too much like mine)

one fun consequence of the CSM-1B voice model accepting audio context is that you can create pretty diverse voices by cloning your own shitty impression of your desired voice/accent

is there a reason that "intellectual" forums (e.g. HN, EJMR) are way more toxic than normal niche communities (e.g. subreddits)

@posedscaredcity @vogentai sure thing, just Dmd!

@jagath_vyth @vogentai yeah i'd love to - im pretty busy today but let's definitely meet this week! can you dm me your email? or someone on your team? thanks :)

Today we're launching Vogent Voicelab: an optimized API to run top open-source voice models, like Sesame's CSM-1B, Dia, Orpheus, and more.

@posedscaredcity @vogentai Hey! Lemme dm you and we can hop on a call?

@jagath_vyth @vogentai we were about to sign a deal with 11labs but im gonna delay that to see if you folks can get there!

@posedscaredcity @vogentai Hey thanks for subscribing! The concurrency is for tts requests, but let me bump up your concurrent dials as well. Thanks for pinging!

@_bprimal_ yeah! you can use the vogent agent builder for that, or you can use our streaming api. check out docs.vogent.ai

@jagath_vyth Can you stream for a live conversation?

made an app to generate theo von podcasts on demand

@qw3rtman 🙏🙏🙏