High-performance features for ultra-low latency voice
Stream real-time voice with ultra-low latency, developer-friendly APIs, and reliability built for enterprise grade AI calling.

<20 ms
Ultra-low latency
99.99%
Session success & Uptime guarantee
24 kHz
HD audio streaming quality (Beta)
0—M
Elasticity >> scale 0 to millions calls
Powering 300 mn+ calls. Trusted by enterprises.




































Experience AgentStream in minutes
Create an account
Navigate to sandbox account and go to app bazaar
Select the voicebot applet and enter the websocket URL
Get your number
Connect your AI bot via websocket

Start using AgentStream now
Building a voicebot? You'll need to connect it to PSTN, WhatsApp, or in-app channels.
For WebRTC users, AgentStream pairs with our Digital Voice SDK for seamless voice delivery.

Do more with Exotel AgentStream
AI-powered inbound & outbound calls with real-time voice streaming.

Collections
Build voice bots to boost recovery, powered by AgentStream.

Sales
Accelerate lead qualification by building AI voice agents streaming on AgentStream.

Support
Boost first-call resolution by building responsive support bots running on AgentStream.

Service
Build always-on voice support bots streaming via AgentStream to handle routine queries.
Reliable infrastructure
Compliant with TRAI & PCI DSS
Ensure secure, compliant operations with TRAI and PCI DSS standards
UL-VNO license holder
Enable seamless IP–PSTN intermixing with a compliant UL-VNO backbone
25B+ Calls handled annually
Scale effortlessly with infrastructure proven to handle 25B+ calls yearly
Supported across 11 circles
Expand nationwide with reliable coverage across all 11 telecom circles
What is voice streaming?


Voice streaming (AgentStream) is the real-time transmission of live voice data during a call to AI engines, supervisors, or external systems.
Unlike static calls, voice streaming enables mid-call intelligence, live transcription, real-time assist, bot actions, multilingual support, and supervisor intervention while conversations are still in progress.
With voice streaming, you can monitor, analyse, automate, and enhance conversations as they happen, not after they end.
Frequently asked questions
What is voice streaming and how is it different from traditional voice calling?
Voice streaming enables real-time transmission of voice data during a call, allowing AI engines, analytics systems, and supervisors to access and act on live conversations — unlike traditional calls that are isolated and static.
Do i need to have technical expertise to set up voice AI bots on Exotel?
Not at all. Exotel provides no-code and low-code options to build voice bot flows, and offers easy integrations with CRMs, knowledge bases, and AI engines to help you go live in minutes.
Can i integrate my own AI/LLM engine with Exotel’s voice streaming?
Yes. Exotel’s platform supports flexible APIs and real-time voice stream delivery that can connect to your custom-built LLMs or third-party AI/NLU engines for mid-call automation and analysis.
What kind of calls can be supported with voice streaming?
Exotel supports agent-to-customer calls, bot-led calls, hybrid handoffs, IVRs with live monitoring, outbound reminders, click-to-call, multilingual conversations, and even machine-to-machine (bot ↔ bot) voice interactions.
Is the platform scalable for large-scale use cases like campaigns or collections?
Absolutely. Exotel’s infrastructure is enterprise-grade, designed to handle millions of concurrent calls with real-time intelligence, routing, and automation across regions.
How secure and compliant is Exotel for regulated industries?
Exotel is UL VNO-certified in India and follows stringent security protocols. It is compliant with local telecom regulations and data privacy standards, making it ideal for industries like banking, insurance, healthcare, and government.
What analytics and reporting capabilities are available?
With voice streaming, you get access to real-time transcription, sentiment analysis, keyword triggers, agent scoring (via CQA), and post-call summaries — all accessible via dashboards or APIs.
Can i use Exotel for multilingual or translated calls?
Yes. Exotel’s voice streaming supports real-time language detection, translation, and multilingual voicebot deployment for regional and international use cases.
What kind of bots are supported?
Any bot that can receive audio over WebSocket.
What audio formats are supported?
PCM 16-bit 16kHz is default. PCM 24kHz available via up/down sampling.
How quickly can i go live?
In under 10 minutes using our setup flow and GitHub code samples.
Can i handle retries and disconnects automatically?
Yes, Exotel AgentStream supports bot-led retries and passthru fallback.
Is it compliant with telecom regulations?
Yes, Exotel AgentStream is PCI DSS, DLT, TRAI, and UL-VNO compliant.
