Voice agents that pick up when your team can't.
Talknex answers inbound calls, qualifies callers, and books appointments straight into your calendar. Responses land in under 600ms, with natural prosody and clean interruption handling.
Currently in private beta
Built for clinics, brokerages, and support teams that can't afford to miss a call. We're onboarding a small group of early teams — request access to join.
Capabilities
A complete platform — not a wrapper
Four pillars, one stack. Build, deploy, and scale autonomous voice agents without stitching together five vendors.
Real-time voice
Conversations indistinguishable from a human.
Streaming Deepgram STT, GPT-4o reasoning, and ElevenLabs Turbo TTS pipelined together for sub-600ms responses with natural prosody and graceful interruption.
WhatsApp + SIP
Reach customers on every channel they answer.
One agent, every channel: PSTN voice via Twilio, WhatsApp Business, and direct SIP trunking with carrier portability. Bring your own number or buy in-app.
Multi-agent system
Spin up an entire workforce, not just an IVR.
Compose specialised agents with shared memory, hand off mid-call, and orchestrate workflows across departments with deterministic routing rules.
Knowledge-grounded
Answers backed by your real source of truth.
Drop in PDFs, crawl your site, or sync Notion. pgvector RAG keeps responses grounded — no hallucinations, with citations exposed in transcripts.
Live demo
Hear our agents in the wild
Pick a persona to preview. Sign up to talk to a live agent on your number, on your data, in your voice.
How it works
Live in three steps
From signup to taking real calls in less than 15 minutes — no engineers required.
Create your agent
Configure persona, voice, system prompt, and connect your knowledge base in minutes.
Connect a number
Provision a Twilio number or plug in WhatsApp / SIP with one click.
Go live 24/7
Your agent answers, books, transfers, and reports — while you sleep.
For developers
An API your team will actually enjoy.
REST + WebSocket. Type-safe SDKs for Node and Python. Webhooks for every call event. Spin up an agent, place calls, and stream transcripts with one import.
- Streaming transcripts over WebSocket
- Webhooks for every call event
- Function calling for tools
- Idempotent retries built-in
# Place a call with a Talknex agent
curl -X POST https://api.talknex.ai/v1/calls \
-H "Authorization: Bearer $TALKNEX_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"agentId": "agent_aria_8f3c",
"to": "+14155550192",
"metadata": { "leadId": "lead_4421" }
}'
Performance
Built for low-latency conversations
Streaming STT, GPT-4o, and Turbo TTS are pipelined to keep median round-trip latency around 580ms — under our 600ms target.
Median round-trip latency, measured on our own pipeline
Deepgram emits transcript deltas every ~120ms — we feed GPT before the caller finishes speaking.
Predictive intent classification skips full inference for short responses like confirmations.
Audio chunks stream back before the LLM finishes, so callers hear the first words while the response is still being generated.
Use cases
Built for every industry
From clinics to brokerages, teams use Talknex to handle the calls they used to miss.
Who it's for
Built for teams that can't miss a call
Talknex is currently in private beta. We're working closely with a small group of clinics, brokerages, and support teams to get the details right before opening up.
Answer patient calls, take intake details, and book or reschedule appointments after hours.
Qualify inbound leads the moment they call and schedule tours without playing phone tag.
Resolve common tier-1 questions instantly and warm-transfer anything that needs a human.
What teams are saying
Built for teams that live on the phone
“We went from missing almost half our after-hours calls to zero. Talknex books appointments straight into our calendar, and the front desk finally has room to breathe.”
“It qualifies every inbound lead the second the phone rings, so our agents only spend time with people who are actually ready to tour. Like an SDR that never sleeps.”
“Tier-1 ticket volume dropped by about a third in the first month. Most callers can't tell it isn't a person until we tell them.”
Why Talknex
One stack, batteries included
The capabilities you need to run production voice agents — without stitching together five vendors.
| Capability | Talknex | What it means |
|---|---|---|
| Median round-trip latency | ~580ms | Under our 600ms target on identical utterances. |
| Voices (ElevenLabs) | Hundreds | Plus custom voice cloning on paid plans. |
| Languages | 30+ | Speech-to-text and text-to-speech across major locales. |
| Calendar booking | Built-in | book_appointment writes straight to Google or Outlook. |
| Telephony channels | Voice, WhatsApp, SIP | Via Twilio — bring your own number or buy in-app. |
| Knowledge base (RAG) | Included | pgvector embeddings keep answers grounded in your docs. |
| Human handoff | Built-in | transfer_to_human warm-transfers the live call. |
| Streaming SDK + webhooks | Included | REST + WebSocket transcripts and a webhook for every event. |
What you get
Built on a real-time stack
The product facts that matter when you put a voice agent on a live phone line.
Documentation
Everything you need to ship an agent
Security & compliance
Enterprise-grade from day one
Pricing
Pay only for what you use
No fixed tiers. Pick a billing mode and pay by the minute — bring your own keys, or let us run the whole stack.
Bring Your Own Keys
Plug in your own provider keys. No monthly base. Your first concurrent line is free, then $20 per additional line.
- $0 monthly base
- $0.05 per minute
- First concurrent line free
- $20 per extra line
Fully Managed
We run the infrastructure end to end — no keys to manage. $99/mo base plus $0.12 per minute and $45 per concurrent line.
- $99 monthly base
- $0.12 per minute
- $45 per concurrent line
- Zero provider keys to manage
Integrations
Plays nicely with your stack
Native integrations for the tools teams already use — and a webhook for everything else.
FAQ
Got questions?
Answers to the things teams ask before they ship their first agent.
How quickly can I deploy a voice agent?
What languages and voices are supported?
Will my agent sound like a robot?
Can agents book appointments and transfer to humans?
How does pricing work?
Is my data secure?
Do you support WhatsApp and SIP?
How am I billed for usage?
Voice AI moves fast. Keep up.
One short email a month — product updates, benchmarks, and the occasional teardown.
Stop missing calls.
Start scaling conversations.
Join the teams using Talknex to answer every call, qualify every lead, and book every appointment — automatically.