Real-time voice (API)

Teach your app to talk

Your user taps a button and speaks to the same agent that answers your phone — its knowledge, its manner, its permissions. No numbers, no phone lines: just a microphone and the internet.

  • WebRTC
  • x0.5/min
  • SDK for any platform

A voice that lives beyond the handset

No awkward pauses

The agent hears and replies almost instantly — to the user it feels like ordinary conversation, not “please hold while I process that”.

Runs on just about anything

JavaScript, Swift, Kotlin, Flutter, Unity, Python — a website, a mobile app, a kiosk or any device with a microphone.

Connected in one request

Ask the platform to open a voice session — and the device is already talking. The SDK handles all the plumbing.

One agent for every channel

Set Maia up for phone calls and she answers in your app too: shared knowledge, scripts and permissions.

Where it fits

The real-time API is how you put live voice into your own product. The text chat widget on your site stays text-only; making the web and apps speak is exactly what this channel is for.

  • A voice assistant in your mobile or desktop app — the customer talks without leaving the interface
  • An assistant on your site, on top of your own design: visitors ask out loud instead of digging through menus
  • Kiosks, terminals, game engines, smart devices — any hardware with a microphone
  • Fun fact: the in-cabinet voice test in MAIA runs on this very channel

Why give your product a voice

A “call support” button throws the customer out of your app. Live voice inside keeps them with you.

  • Account area: the customer asks about a plan or an order out loud and gets the answer in seconds
  • Onboarding: the agent answers “where do I find…” right during the first session with the product
  • Driving, working out, hands full — voice steps in where typing is a chore
  • Update the agent's knowledge and the app answers the new way — no release, no app-store review

How to connect

  • Request a voice session with your MAIA key
  • Plug the SDK into your platform and hand it the microphone
  • Done: the user is chatting with the agent — knowledge base and actions already in place

How it differs from telephony

  • No phone number and no carrier — the conversation travels over the internet
  • One and the same agent: trained for calls, works here too
  • A minute costs half the phone rate: a x0.5 multiplier
  • Perfect where a carrier call is overkill — the customer is already inside your product
FAQ

Frequently asked

Is this a ready-made voice widget for my site?
No — it sits one level lower: an SDK your team uses to build live voice into the product. MAIA's ready-made site chat widget talks in text.
Which platforms are supported?
Anything that runs the SDK: JavaScript, Swift, Kotlin, Flutter, Unity, Python and more — from a website to a smart speaker.
Is it the same agent as on the phone?
Literally the same one: shared knowledge, scripts and actions. Only the transport changes — a device microphone instead of a phone line.
How much does it cost?
Half the price of a phone call: a x0.5 multiplier on the platform base rate, which starts at 5 ₽/min.
Do I need a developer?
Yes — embedding is a job for your developer: plug in the SDK and pass it the microphone. Everything else is already set up: the agent, knowledge and actions come from your MAIA project.

Make your product talk

Tell us where the voice should live — app, website or device — and we'll put together a connection example for your platform.

Trial balance after sign-up — call your agent right from the browser. No card required. Book a demo →