korinne retweeted
Your voice agent nailed the demo. Can it handle the next thousand conversations?
Join Cloudflare’s @korinne_dev and Coval CEO @bnicholehopkins for a fireside chat on voice agent latency, conversation state, and testing in production.
Tuesday, Nov 3 @ Cloudflare SF
5:30 - 8 pm
Register below
x.com/bnicholehopkin…
Join Cloudflare’s @korinne_dev and Coval CEO @bnicholehopkins for a fireside chat on voice agent latency, conversation state, and testing in production.
Tuesday, Nov 3 @ Cloudflare SF
5:30 - 8 pm
Register below
x.com/bnicholehopkin…
On November 3, we’re hosting a fireside chat and live demo with @Cloudflare at their SF office. Would love to see you there!
Register: luma.com/iq4hna2x
4
3
37
4,616
korinne retweeted
🎨 Custom pages for @Cloudflare Access are here.
Use HTML + Liquid to brand login, access-denied, and CLI approval pages. Give blocked users clear next steps, create custom branding, and make it easier on your users!
Check out the new demo denied page I cooked up 🤩
🔗 developers.cloudflare.com/changelog/post…
Use HTML + Liquid to brand login, access-denied, and CLI approval pages. Give blocked users clear next steps, create custom branding, and make it easier on your users!
Check out the new demo denied page I cooked up 🤩
🔗 developers.cloudflare.com/changelog/post…
33
40
400
46,460
Very excited to sit down with @bnicholehopkins to talk about voice AI. Come by the SF office, and show me what you're building!
On November 3, we’re hosting a fireside chat and live demo with @Cloudflare at their SF office. Would love to see you there!
Register: luma.com/iq4hna2x
3
3
13
1,690
korinne retweeted
hanging out with @korinne_dev at the Project Build booth 🤓 — don’t miss out on Korinne’s talk on Agents that Talk 🗣️
meet your moots folks! 🙂↕️🤝
meet your moots folks! 🙂↕️🤝
A WILD @Nateemerson APPEARS!
lovely meeting you dude 🫂
meet your moots folks! 🙂↕️ 🤝
ft my new best friend @samhuckaby
1
2
29
1,420
korinne retweeted
phonellm on workers ai is the gift i did not know i needed
Be sure to use the prompt in our changelog for PhoneLLM Mini to use Pipecat's newest model
developers.cloudflare.com/changelog/post…
0
2
7
469
korinne retweeted
@korinne_dev @michellechen The biggest single request we got after the PhoneLLM Alpha launch last month was for an endpoint priced per token. @korinne_dev and @Cloudflare delivered!
1
1
4
207
Did you know you can run a Pipecat agent on Python Workers? We didn't either until we made this:
github.com/cloudflare/rea…
It uses WebRTC for transport thanks to Cloudflare Realtime SFU. And I also added some nice keyboard sounds for when the agent looks things up.
And yes, I what's one more Matt at Cloudflare?
github.com/cloudflare/rea…
It uses WebRTC for transport thanks to Cloudflare Realtime SFU. And I also added some nice keyboard sounds for when the agent looks things up.
And yes, I what's one more Matt at Cloudflare?
3
4
28
4,123
Be sure to use the prompt in our changelog for PhoneLLM Mini to use Pipecat's newest model
developers.cloudflare.com/changelog/post…
developers.cloudflare.com/changelog/post…
0
0
3
825
Very excited partner with the Pipecat team to get PhoneLLM Mini on Workers AI.
Congratulations @kwindla and team - love using Pipecat and this model is fantastic. And huge thank you to @michellechen and the Workers AI team for making everything come together.
Congratulations @kwindla and team - love using Pipecat and this model is fantastic. And huge thank you to @michellechen and the Workers AI team for making everything come together.
Fast is better than slow.
We've got two new PhoneLLM models optimized for conversational voice AI.
PhoneLLM Mini (30B) — When you need the lowest possible latency (sub-100ms TTFAT), smallest model size, and lowest cost. Good multi-turn conversation behavior and tool-calling t can run >100 concurrent agents on a B200 and fits on an RTX 5090.
PhoneLLM Max (120B) — A "microthinking" model for complex conversational tasks. Sub-200ms TTFAT. SOTA task completion performance at 1/10th the latency and cost of models with similar capabilities.
Both of these models are full-weights fine-tunes of @NVIDIAAI Nemotron models. Mini is based on Nemotron 3 Nano. Max is based on the brand new Nemotron 3.5 Super VL.
Both of these models are available as open weights on HuggingFace, and from four inference partners: AWS, Baseten, Cloudflare, and Modal.
2
6
22
2,151
In case you missed, Stream Live WebRTC is now GA!
Stream Live is built on our Realtime WebRTC SFU, but designed for broadcasting 1-to-many .
Because it's built for ultra-low latency viewing, it's great for live auctions/betting.
developers.cloudflare.com/changelog/post…
Stream Live is built on our Realtime WebRTC SFU, but designed for broadcasting 1-to-many .
Because it's built for ultra-low latency viewing, it's great for live auctions/betting.
developers.cloudflare.com/changelog/post…
1
4
25
1,133
korinne retweeted
in honor of 20 year 30 rock anniversary + clef-omni release, decided to build a little experiment with workers + @Cloudflare realtimekit
"jenna's side... i'm not hearing it liz"
try it yourself 👉
clef-realtime.rita.workers.dev
"jenna's side... i'm not hearing it liz"
try it yourself 👉
clef-realtime.rita.workers.dev
8
13
150
12,446
korinne retweeted
bring your choice of harness (think, pi, and now opencode)
run it on durable objects (the brains)
use container sandboxes to do work (the hands)
plus the rest of the platform (browser, email, voice, mcp)
run it on durable objects (the brains)
use container sandboxes to do work (the hands)
plus the rest of the platform (browser, email, voice, mcp)
opencode 2.0 runs in durable objects
npm i agents@latest @opencode/sdk@2 @opencode/plugin@2
developers.cloudflare.com/changelog/post…
1
4
47
2,606
korinne retweeted
couldn't leave for the weekend on a clefhanger
clef-omni for audio, video, images + cloudflare realtime / media suite will be a very fun combo
clef-omni for audio, video, images + cloudflare realtime / media suite will be a very fun combo
we've got more 𝚌𝚕𝚎𝚏 for you:
• new 𝚌𝚕𝚎𝚏-𝚘𝚖𝚗𝚒 - takes audio, video, image, and text
• 𝚌𝚕𝚎𝚏-𝚏𝚕𝚊𝚜𝚑 is now cheaper than jev
• 𝚌𝚕𝚎𝚏 is now ~2x faster
blog.cloudflare.com/clef-faster-ch…
8
7
105
4,195
RT @dok2001: Clef-omni is out. We add audio and video input, plus text and images.
@Cloudflare teams are using it to find PII, detect mal…
@Cloudflare teams are using it to find PII, detect mal…
0
9
0
3
korinne retweeted
we've got more 𝚌𝚕𝚎𝚏 for you:
• new 𝚌𝚕𝚎𝚏-𝚘𝚖𝚗𝚒 - takes audio, video, image, and text
• 𝚌𝚕𝚎𝚏-𝚏𝚕𝚊𝚜𝚑 is now cheaper than jev
• 𝚌𝚕𝚎𝚏 is now ~2x faster
blog.cloudflare.com/clef-faster-ch…
• new 𝚌𝚕𝚎𝚏-𝚘𝚖𝚗𝚒 - takes audio, video, image, and text
• 𝚌𝚕𝚎𝚏-𝚏𝚕𝚊𝚜𝚑 is now cheaper than jev
• 𝚌𝚕𝚎𝚏 is now ~2x faster
blog.cloudflare.com/clef-faster-ch…
48
103
877
93,912
I need @Cloudflare to put a PoP in the transbay tube. I know that’s not how cell service works but I need someone to take responsibility
1
1
12
744
The WebSocket adapter for our WebRTC infrastructure (SFU) is now generally available.
This makes it so that any live audio or video taking place on Cloudflare's SFU can be sent to a WebSocket endpoint to be processed.
developers.cloudflare.com/changelog/post…
This makes it so that any live audio or video taking place on Cloudflare's SFU can be sent to a WebSocket endpoint to be processed.
developers.cloudflare.com/changelog/post…
4
8
39
3,443
We use the WebSocket adapter in RealtimeKit to handle our managed transcriptions service: developers.cloudflare.com/realtime/realt….
This means our engineers learn how to improve our own primitives by building higher level products on top, which has been incredible.
This means our engineers learn how to improve our own primitives by building higher level products on top, which has been incredible.
0
0
2
219