Raghu | YANVI

Building YANVI - an AI desk companion running @NousResearch Hermes agent with a physical body, 13 emotions, and full Mac control. She sees your screen, tracks your face, and pushes back when your ideas are bad. Solo build, 14 days, 20K lines of code. Would love to demo live for you.

@tom_doerr Thanks for sharing.

Automates KiCAD panelization and manufacturing data export github.com/yaqwsx/KiKit

Refreshing my inbox at 1 AM on a Sunday awaiting @fdotinc canopy results.. @hthieblot 🤞

someone one fine night looked up at the sky and wondered what was up there, just about 200 years ago. and now we're here - circling the moon like it's a bike ride. I'm just happy I'm alive when 4 humans flew ~400,000 kms from Earth, broke a 56-year-old record set by Apollo 13, witnessed a solar eclipse from behind the Moon for 57 minutes, and splashed down safely off San Diego. what a time to be alive.

@cottagetwt @billieroan Do they atleast have insurance in Millions? 🤔

@billieroan i wonder how much did the 4 astronauts make from this mission?

went to nasa’s career page. let out the loudest laugh known to man. closed the tab. very humbling.

This looks perfect for embodied AI! I'm building YANVI - a physical AI desk companion robot that runs fully local with Hermes agent. Strong real-time AEC + far-field voice is exactly what we need. Already using previous ReSpeaker v2.0 XVF-3000 - this Flex upgrade would be huge. Any chance for a review/sample unit? Would love to test it with Yanvi and share the results! Thanks.

🤖reSpeaker Flex is officially here! Performing #AEC at the Real Time! 🔗 seeedstudio.com/reSpeaker-Flex… Meet the new reSpeaker Flex 4-Mic Array, built for embedded AI applications like #Robotics and smart #DigitalSignage. 🎶Music plays continuously through a connected speaker, while reSpeaker Flex still captures the wake word clearly and responds accurately. Even with strong audio feedback, it delivers reliable far-field voice pickup and smooth voice interaction in real-world environments! What do u want to build with reSpeaker Flex? Let us know in the comments!😲 #SeeedStudio #reSpeaker #reSpeakerFlex #MicrophoneArray #XMOS #VoiceAI #SoundAI #EdgeAI #AIoT #EmbeddedAI #EmbodiedAI #XMOS #Robot #Robotic #DigitalSignage #SmartSignage #VoiceInterface #SmartEarforEmbodiedAI #TheAIHardwarePartner

@MartinGTobias Neurofeel is a model that detects emotions from statements. It has been extremely helpful in my Yanvi robot project. I seriously hope more such niche models are developed. Ref: huggingface.co/boltuix/NeuroF…

does anyone believe local task specific SLMs will have a place in a world where general LLMS are battling on costs and improving at a rapid pace?

@sixxella_moon @Scobleizer @NousResearch Yeah for many! I'm working on Yanvi's expressions and servo movements to make her feel non-robotic.. Is there any specific expression or animation sequence that you still cherish? I can try and add that to her. ❤️

@HeyYanvi @Scobleizer @NousResearch Reminds me of Johnny 5!! Short Circuit was one of my favorite movies as a kid

Now that my AI is reading X for me I am going to pour a lot more effort into interviewing founders again who are starting companies. You one? Drop me a line.

@Tripleace_Ade Learning PCB design as part of my robot project. Let's learn together 💪

Hey Algorithm, let those interested in PCB design and Drones see this post. I'm trying to gain new followers and get more mentors.

@Scobleizer @NousResearch Really? Sure.

@HeyYanvi @NousResearch Ah cool. Will do a video call then. Let’s take it into DMs

@Scobleizer @NousResearch Thank you. I'm in Toronto, Canada

@HeyYanvi @NousResearch Sweet! Where are you located?

@Scobleizer @NousResearch Thank you. I'm in Toronto, Canada

Building YANVI - an AI desk companion running @NousResearch Hermes agent with a physical body, 13 emotions, and full Mac control. She sees your screen, tracks your face, and pushes back when your ideas are bad. Solo build, 14 days, 20K lines of code. Would love to demo live for you.

@jasonkneen @cursorbuddyapp @FarzaTV This is amazing. Love that you exposed it as an MCP server. Also animations with Bezier Flights + Spring physics is chef's kiss.. Thank you.

Introducing @cursorbuddyapp -- inspired by clicky by @FarzaTV (thank you sir) and written from the ground up to be multi-platform ready, it provides a complete cursor agent that knows what you're doing, and comes as: Feature list now and soon - desktop - cli - web injection script (script include) - MCP - full voice - computer use - proper tooling - plugins and custom extensions and tools - themes - a playground - local options - detects screenshots - *walks you through learning - *time coded voice to text to get natural streaming - *multiple "canvas" views for interacting - *image support - *draws and annotates your screen - *interacts / available with agents to help out - *multiple "buddies" with characterisations and skills - *multi-player with other remote buddies* *roadmap will post the repo in this thread shortly, all free, all MIT license.

Also fixed a high-severity Dependabot alert (CVE-2026-32285) 🔒 Updated transitive dependencies to patch a DoS vulnerability in jsonparser. • mark3labs/mcp-go v0.43.2 → v0.47.1 • mailru/easyjson v0.7.7 → v0.9.2 No more vulnerable deps in the tree. Tests pass, binary works. github.com/Pilan-AI/mnemo…

mnemo just hit 24 stars & 5 forks! ⭐ Big thanks to the community for the support. We just shipped a security fix with v1.3.4 Removed unused HTTP proxy code that had an auth vulnerability (issue #5). The proxy was experimental & never wired into the CLI. mnemo's architecture (MCP server + hooks) doesn't need HTTP interception anyway. Cleaner codebase, tighter security. ✅ github.com/Pilan-AI/mnemo…

@Scobleizer @openclaw @NousResearch @theemozilla Exciting! I've been working on a 2-layer system where a hermes agent on a Pi-5 delegates tasks to another hermes agent on the Mac for complex tasks. This would enable multi device agent orchestration easier. Any thoughts or inputs on this pls?

You have heard of @openclaw competitor from @NousResearch called “Hermes.” Tomorrow at 4 pm we will get nerdy with @theemozilla. Live. I will get people up who asks questions here first. twitter.com/i/spaces/1DGle…

I'm building an AI desk companion robot 'Yanvi' and primarily using Voxtral TTS for her voice.. The Voxtral TTS voice sounds so natural with pauses and intonations and there's also subtle breathing sounds.. All of this make Yanvi feel more like a companion than a robot. Amazing @MistralAI

🎙️Designing a speech-to-speech assistant - Voxtral Transcribe 2 for STT + diarization - Mistral Small 4 for agentic reasoning & efficiency - Voxtral TTS for realistic speech synthesis Build a speech-to-speech assistant with web search access in 150 lines of code.

@TechCrunch What a great opportunity! I'll apply with my best effort..

Startup Battlefield 200 applications open: A chance for VC access, TechCrunch coverage, and $100K techcrunch.com/2026/04/06/sta…

@zwiebelhelm @fdotinc Love the energy.. Wish you the best from a fellow applicant!

Demo/PoC for Chai Computer @fdotinc