Co-Founder @livekit. Entrepreneur and engineer. I like computers and believe in hard money. #Bitcoin
The Internets
- Tweets 1,609
- Following 828
- Followers 2,094
- Likes 9,088
David Zhao retweeted
A demo of a practical use case for Jev: routing customer inquiries on a live call center line
Fun voice AI project built with @livekit
It's really incredible to see how quickly Jev can act on a live call and bring in real-time AI judgment on everything the caller says
14
3
38
5,393
David Zhao retweeted
Gemini 3.8 Flash-Lite TTS is nice to talk to.
What do you think?
Is your voice agent moody? It should be!
With Gemini 3.8 Flash-Lite TTS, you can direct its mood or let the agent decide for itself. Cheerful, furious, wistful, apologetic, it nails all of them and sounds great.
Give it a try on LiveKit: livek.it/obFIyE5
1
3
26
3,080
Enterprise agents need secure access to internal systems.
Private links on LiveKit Cloud open a fully-manage, encrypted tunnel straight into your VPC without needing public IPs, inbound firewall rules, or VPN.
Live today in the US and EU.
Learn more > livekit.com/blog/introducing…
2
3
34
10,091
David Zhao retweeted
Today we're opening up Nebula Alpha to everyone.
We imagined team communications where agents are a first class citizen.
They work in channels, talk on calls and can see your screen.
All at real time speed.
It's time for every team to become agentic.
Nebula
16
24
117
23,932
Your whole voice stack, free for a year on LiveKit.
$10K in Cloud credits + a year of Scale + $7K in partner inference credits ($1K each: @googlegemma, @DeepgramAI, @inworld , @rimelabs , @FishAudio, @AssemblyAI, @GradiumAI.)
One workspace. No provider keys.
Apply now: livekit.com/startups
10
13
124
10,430
We power real-time infra for robots in warehouses, in the air, in your home. We want to meet the people building them in person!
Robotics Happy Hour w/ @dimensionalos, #SFTechWeek
Tue Oct 6 · 5:30pm · SF
RSVP here→ partiful.com/e/u610QKw40q4lh…
4
3
16
1,236
Gradium TTS is now available on LiveKit Inference, free until Oct. 9.
Gradium TTS is designed for conversational use cases, with voice cloning, low latency (below 250ms TTFA), and robust pronunciation on difficult cases.
Select @GradiumAI as your TTS provider in LiveKit Inference, no additional account or API key required: docs.livekit.io/agents/model…
7
7
61
7,192
@livekit is one of the easiest ways to build a voice agent, and if you've been using another voice vendor (that rhymes with shmelevendrabs) on LiveKit's Inference product, you're eligible for free TTS from Rime through the month of September.
Just make the switch in LiveKit to any Rime voice and start leveling up your voice agent game.
rime.ai/livekit-switch-to-ri…
1
5
2,729
SFist writer needs a math lesson on how far 13 ft is before writing headlines.
A father in Noe Valley says a Waymo robotaxi stopped within 13 feet of his two-year-old daughter when she fell in the crosswalk, nearly running her over, but Waymo claims the car wouldn’t have hit her.
sfist.com/2026/09/01/father-…
2
11
1,273
the possibilities are endless once you can generate video faster than real time.
Introducing fal.live
A new platform for infinite, interactive AI livestreams. Pick a channel, prompt what happens next, and watch it generate in real time.
You aren't just watching the show. You're directing it.
1
17
1,793
The best PII redaction system for voice AI.
Protect sensitive data without giving up voice agent observability.
LiveKit PII Redaction automatically removes personal information from transcripts and recordings before it is stored. It is included with LiveKit Agent Observability at no additional cost.
Read more: bit.ly/4gJ1ex7
18
1,435
LiveKit Inference is fast
Just tried it. Needless to say the voice agent I tested with it had exceptionally low latency. Seriously Impressive.
E2E latency at 994 ms feels amazingly fast in real conversation. The pipeline showed 250 ms TTS TTFB. Again excellent number.
Transcription delay of 483 ms and interruption detection at 248 ms make it feel very responsive and natural.
Overall, it's one of the quickest and most accurate voice experiences I've tried. It handled everything I threw at it with high precision and almost no noticeable lag.
Great work @livekit!
2
16
8,911
Frontier models keep getting smarter…and slower. This is bad for voice agents, where every millisecond before the first word counts.
So we optimized for speed. Introducing Gemma 4 31B on LiveKit Inference:
🗣️ 381ms to first sentence — 2x+ faster than the next model
✅ 88% task completion (see the criteria in reply)
💰 $1.20 / 1M output tokens
Fast enough for real conversation, smart enough for effective agents.
10
19
116
18,059
Huge congrats to @lina_colucci and the @LemonSliceAI team on bringing Teddy Roosevelt back to life at the new Theodore Roosevelt Presidential Library!
Super cool to see what you built. Thrilled our tech could play a small part in it.
President Trump asks AI President Roosevelt…“Do you consider the Panama Canal your greatest achievement?”
2
11
1,083
David Zhao retweeted
Self-hosted voice assistant built on LiveKit Agents
github.com/CoreWorxLab/CAAL
7
33
3,694
End-of-turn detection is solved with our SOTA v1 model.
It fuses an audio encoder directly into the LLM backbone, so the model can use both audio cues and semantic understanding together. It's the fastest and most accurate turn detection model we've tested.
We shipped LiveKit Turn Detector v1.
Instead of reading transcripts, it listens to speech directly, combining semantic and acoustic cues into one end-of-turn prediction.
The result: high accuracy, low latency—the best model we tested across 14 languages.
Available on LiveKit Cloud.
1
11
978
You can now run an agent in the cloud that remotely controls a commodity robot (without any specialized inference hw).
The team used this stack last month to win a @spc hackathon.
Operating a robot over the internet means camera frames and joint state arrive at different times, so your observations drift and training data gets misaligned.
LiveKit Portal fuses them back together with the same code, whether the robot's in the next room or another continent.
2
13
1,137