ClawdTalk: Voice Calls for ClawdBots
Hey HN, We built ClawdTalk to let AI agents operate over live phone calls. Most agents today live in chat windows. The moment you try to use them over voice, things break: latency matters, interruptions happen, and the agent has to execute tools while the conversation is still live. ClawdTalk connects your Clawdbot to the phone network. The agent gets a real phone number. It can make and receive calls and run the same tools it uses in chat, but under real-time voice constraints. One of the reasons this is hard is infrastructure. Many voice stacks stitch together separate telephony, speech,…
In plain words
ClawdTalk enables AI agents to operate over live phone calls by connecting them to the phone network with a real phone number. Agents can make and receive calls while executing the same tools they use in chat, optimized for real-time voice constraints with latency under three seconds. Built by running the full voice path—telephony, speech-to-text, and text-to-speech—on integrated infrastructure rather than stitching separate services together, it addresses the technical challenges of voice-based AI interactions that break down in traditional chat-based systems.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hey HN, We built ClawdTalk to let AI agents operate over live phone calls. Most agents today live in chat windows. The moment you try to use them over voice, things break: latency matters, interruptions happen, and the agent has to execute tools while the conversation is still live. ClawdTalk connects your Clawdbot to the phone network. The agent gets a real phone number. It can make and receive calls and run the same tools it uses in chat, but under real-time voice constraints. One of the reasons this is hard is infrastructure. Many voice stacks stitch together separate telephony, speech, and model APIs. Each hop adds latency, and people report 8–30 second round trips. We got it under ~3 seconds by running the full voice path ourselves. Telnyx (my employer) is a telecom carrier, and we run PSTN, STT, and TTS on our infrastructure. No middlemen. How it works: 1. Connect your OpenClaw agent to ClawdTalk 2. We provision a phone number 3. Inbound/outbound calls route directly to the agent 4. The agent executes tools mid-conversation Limitations: 1. Latency still depends on your LLM (we control voice, not inference) 2. US numbers only for now (international coming) 3. Not a new agent framework (OpenClaw only today) Demo number: +1-301-MYCLAWD (692-5293) (call to talk to the agent) Happy to answer questions about the architecture or telephony side.
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
