Tone is the telephony API for AI agents — provisioning, compliance, transcription, and CRM sync built for the Indian market.
curl -X POST api.usetone.ai/v1/numbers◇ WORKS WITH YOUR STACK
◇ Compliance
Do-Not-Call
Opt-outs and complaints block the dial automatically. Every check recorded before the call is placed.
DLT Templates
Registered DLT headers inserted on every outbound call — zero manual steps.
Calling Hours
Promotional calls blocked outside 09:00–21:00 IST, server-side.
AI Disclosure
"This is an AI call" prepended to every conversation. Required by TRAI. Automatic.
Opt-Out
Opt-out press stored immediately, honoured on all future calls from your account.
Audit Trail
Every call event timestamped and immutable. Ready for regulatory inspection.
Sandbox
Test your full integration without making real calls or spending credits. tone_test_ keys simulate the complete call flow — transcripts, webhooks, compliance logs — all included.
Multilingual
Hindi, Tamil, Telugu, Bengali, Kannada, Malayalam, Marathi, Gujarati, Punjabi, Odia, Assamese, English — auto-detected via Sarvam Saaras v3.
◇ Quick start
Create account, verify business, copy key from dashboard.
POST /v1/api-keysOne POST and your agent owns a dedicated Indian phone number.
POST /v1/numbersReceive transcripts, respond with text. Tone handles the voice plumbing.
POST your-server/webhook◇ Pricing
No monthly commitment.
Start for freeFor teams shipping their first AI calling agent.
Start for freeFor agents making hundreds of calls daily.
Start for free◇ Resources
Measured numbers, the regulation stated plainly, and what automated calling is really used for in India.
What happens between a caller finishing a sentence and an agent answering it: voice activity detection, streaming speech recognition, the language model, speech synthesis, and the four places it usually breaks.
India's telemarketing regulation, read from the gazette text rather than from vendor blogs — the number series, the pre-declaration duty, the calling window, and the rules that do not exist.
A stage-by-stage latency breakdown of a production voice agent running in Mumbai — where every millisecond of a 1,220ms reply goes, and why in-region matters more than model choice.
◇ Get started today
Join the teams building the next generation of Indian AI voice agents on Tone.