LIVE
Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|Google Introduces "Gemini 4 Carbon" as Next Step in Its Model Lineup28/10/26 · Google|DeepMind unveils Genie 4, a new generation of world model27/10/26 · Google DeepMind|Anthropic files IPO prospectus, with risk disclosures taking center stage26/10/26 · Anthropic|Robot policy trained only in simulation matches top real-world systems26/10/26|OpenAI unveils GPT-6.1 Terra, a model built for computer use26/10/26 · OpenAI|Google's AI gives evasive answers about who won the 2020 US election28/10/26 · Google|NVIDIA introduces Nemotron 3.5 Transcribe, a promptable multilingual transcription model28/10/26 · NVIDIA|OpenAI and ServiceNow deepen partnership on enterprise workflows28/10/26 · OpenAI|Failed startups' data finds a second life training AI models28/10/26|Anthropic expands AI for Science program to 20 foundations and research institutions28/10/26 · Anthropic|Anthropic shares early results on using Claude to accelerate drug development27/10/26 · Anthropic|GEPA-Curator: Filtering Failures to Better Train LLM Agents27/10/26|
Toolofficial publisher

Twilio ConversationRelay

Tool : A precise action the AI can call · requires twilio

Real-time phone voice connected to an agent via WebSocket.

The reference telephony building block for a voice agent.

Docs
Overall score
5.9 /10
Security: one caveat, see the card
Utility
7.0
Reliability
6.0
Security
One caveat
Ease
Technical · Technical

Security reflects where data goes, requested permissions and incidents on record at the verification date. It is not a guarantee of zero risk: check the deployment mode and the connections you enable.

What it does, concretely

Building a voice AI agent over the phone requires handling speech recognition, text-to-speech, and real-time connectivity all at once: Twilio ConversationRelay takes care of these three layers so you can focus on your application logic.

In practice, an inbound or outbound call is connected to your server via a secure WebSocket. Twilio transcribes the caller's speech, sends the text to your agent, then plays the agent's response back as synthesized voice directly in the call, all in real time.

Available speech recognition and synthesis engines include Deepgram (Nova 2), Google (Speech-to-Text and Text-to-Speech), Amazon (Polly), and ElevenLabs (Flash 2 and Flash 2.5). You choose the provider based on your naturalness or cost requirements.

This tool is aimed at developers who want to connect an LLM or AI agent to a phone number via an API. It is not suited for no-code use cases or teams looking for a turnkey solution without technical integration.

What it offers

  • ·Real-time WebSocket connection between a Twilio phone call and your application server (paid).
  • ·Automatic speech recognition via Deepgram Nova 2, Google Speech-to-Text, or other supported providers (paid).
  • ·Text-to-speech synthesis via Google Text-to-Speech, Amazon Polly, or ElevenLabs Flash 2 / Flash 2.5 (paid).
  • ·Configurable welcome greeting delivered as soon as the call connects, without an additional round trip.
  • ·Configurable HTTP callback action at the end of the ConversationRelay session to resume the TwiML flow.
  • ·Logging and auditing of inputs and outputs accessible from your own terminal for human review.
  • ·Compatibility with Twilio-published integration guides for OpenAI, AWS, and BentoML.

In practice

Price
Paid
Commercial use
Commercial use: yes
Available on
api
Host
twilio
FR interface
No

Security and data

Rating
Security: one caveat, see the card
Capabilities
readswritesexecutesnetwork
Your data
Data with third parties
Maintenance
Active

Links and repository checked on 25 October 2026 · added on 1 January 2025

How to install Twilio ConversationRelay

1. Create a Twilio account and obtain a Twilio phone number with Voice capabilities enabled. 2. Install the Twilio SDK for Node.js (or your language of choice) and set your API credentials as environment variables. 3. Set up a publicly accessible WebSocket server (wss://) that will receive transcriptions and send back your agent's responses. 4. Write a TwiML webhook that uses the <Connect><ConversationRelay> tag pointing to your WebSocket server URL and defining a welcomeGreeting. 5. Configure your Twilio number to call this webhook on each inbound call, or trigger it from your code for outbound calls. 6. Verify that audio transfers to Twilio and your chosen voice provider comply with your legal obligations before going to production.
TwiML <ConversationRelay>

Official documentation ↗

In the same category

Whole category Automation & agents →