One platform. All real-time workflows.

Answer every call, run every meeting, reach every audience.

AI phone agents, click-to-call on your website, outbound campaigns, voice and vision agents, meetings with AI notes, live events with translation, widgets and SDKs. One account, at infrastructure rates.

Calls and phone agents

Answer every call, and make the ones that matter.

Phone and AI phone agents

An AI agent answers every call. Your team takes over when it matters.

Put an agent on a real number for calls in and out. Callers can cut in and it stops to listen, and summaries go on to your CRM or workflow tool.

  • Inbound and outboundAnswer a support or booking line, or call people back, with the same agent.
  • Human takeoverWatch the live transcript, take the call over with the full context, and hand it back.
  • Your SIP or our numbersUse one of our numbers or bring your own SIP line. Carrier minutes bill at your provider’s rate.
Example phone line
Click-to-call on your websiteCuts call costs

Turn your website into a phone line.

List your business number as a virtual line. Visitors tap to call from your site, in the browser, and your AI agent or your team answers.

  • No carrier minutesWebsite calls need no SIP trunk and no extra lines. They carry MediaSFU media minutes only.
  • Many callers at onceEach click is its own call, so several visitors can call at the same time.
  • AI or a team queueYour agent answers, or your team picks calls up from a queue, with AI hand-over.
Example website line
Outbound campaigns

Call your whole list. An AI agent handles every conversation.

Give the agent a purpose and a script, set the pace and the hours, and follow every outcome as each call ends.

  • Paced batchesUp to 100 contacts per batch. Set simultaneous calls and the gap between them.
  • Time-zone aware schedulesAllowed hours and days per campaign, so nobody gets a call over dinner.
  • Hand off to a personTransfer a call to your team mid-conversation, and see every escalation afterwards.
Example campaign monitor
Voice and vision agents

AI that hears your customers and sees what they show it.

Multimodal voice and vision agents

An agent that hears you, sees your camera and answers in real time.

Put it in your site or app as a voice and camera widget, or give it a face of its own. Start with a hosted agent, then bring your own pipeline.

  • Voice and visionSpeech-to-text, a language model and text-to-speech, plus what the visitor shows the camera, in one conversation.
  • Your models, no markupBring the speech, language and voice providers you trust. They bill you directly.
  • A hosted agent to startTalk to one with no keys of your own, then move the pipeline into your product.
Example web agent
Meetings and live events

Meet, record and go live to any size of audience.

Meetings, AI notes and recording

Meetings that write their own notes.

Rooms people join from a link on any device, with a transcript and an AI summary afterwards.

  • Join from a linkGuests open the invite on a laptop or phone. Nothing to install, no account to make.
  • Run the room your wayScreen share with annotation, whiteboard, chat, breakout rooms, polls and raised hands.
  • Recording and AI notesRecord when you choose, then read the transcript and summary. Recordings leave MediaSFU 72 hours after they finish.
Example meeting with AI notes
Live events, broadcast and translation

Go live to 500K+ viewers in real time, each in their own language.

One stage for speakers and any size of audience, with translated audio and captions so nobody settles for one shared language.

  • One stage, any audience500K+ viewers in real time over WebRTC or WHEP, and HLS for audiences beyond that.
  • Cut cameras to airAdd a phone as a camera, or go live from OBS over WHIP. Record it, with paywall-ready access.
  • 50+ languagesTranslated voice and live captions. Each listener picks the language they hear.
Example live event
Widgets, SDKs and kits

Ship it your way: a widget, your own code or a ready kit.

Widgets and no-code studios

Add calls, meetings and agents to any website, without code.

Prebuilt website experiences you set up in a studio or the dashboard. Domain keys, owner-scoped dashboards and passcode sharing keep them yours.

  • Every kind of widgetCall buttons, meeting joins, AI agents, web agents, and calls and agent dashboards.
  • Widget StudioShape the widget’s behaviour and look without code, then paste one snippet.
  • Vibe StudioPlan a realtime app, review its roles and media, and export a React project.
Example website with a widget
SDKs and APIs

Build your own product on the same rooms, agents and phone lines.

Open-source SDKs for web and mobile, a REST API, and a sandbox to try calls before you write a line.

  • Prebuilt or headlessShip the complete room UI today, or keep the room engine and draw every pixel yourself.
  • Cloud or your own serverRun on MediaSFU Cloud, or point the same SDK at MediaSFU Open on your own infrastructure.
  • Phone and agents on the same roomAdd a SIP number or an AI voice agent to a room through the same account and the same bill.
Example first room in React
Kits and use-case rooms

Start from an app that already works.

Open a working room and play both sides on two devices, or start from a kit with a live demo and a starter you deploy on your own MediaSFU keys.

  • Remote inspectionClaims, warranty and returns, field service

    Customers show the item on their own phone camera. An AI guide collects every required view and hands your inspector a sealed evidence package.

  • Telehealth consultsClinics and teleconsult platforms, wherever their patients are

    Patients consent, wait and join a low-data video consult from a link. Doctors admit them and sign an AI-drafted visit note.

  • Multilingual eventsWebinars, services, conferences and community events

    Speakers talk in their own language. Every listener picks theirs and hears the event translated live.

  • Agency phone agentsAI automation agencies and resellers

    Run AI receptionists, booking lines and lead qualifiers for your clients, with a workspace per client that shows cost and margin. Send every call summary on to your CRM with our n8n node.

  • Video KYC frameworkIndia V-CIP (default) and International ID

    A complete video KYC app your institution runs on its own infrastructure, keeping recordings and customer data there. MediaSFU carries the call.

Trust

Built to be relied on, and open about it.

  • 99.99% uptime SLA

    Watch every service live on the status page.

    System status
  • Encrypted in transit

    TLS for every API call and DTLS-SRTP for every stream. Media is forwarded end to end without decoding, unless you turn on recording, transcription, translation or AI.

    Security
  • Your data, your servers

    Recordings are deleted from MediaSFU 72 hours after they complete. Kits run on your own hosting, and MediaSFU Open runs on your own infrastructure.

    Data and residency
  • Room and audience size

    3,000 people interacting per room, more with large and flexible rooms. 500K+ viewers in real time over WebRTC or WHEP, and HLS for audiences beyond that.

    Size your event
Pricing

Priced like infrastructure. Your providers bill you direct.

Includes media transport, agent runtime and recording. Your AI providers, phone numbers and carrier minutes bill separately and directly.

  • 50,000 free minutes that never expire
  • 20% off from 2.5M minutes a month
  • 40% off from 7.5M minutes a month
Audio in a room or call$0.0001 per minute$1 buys 10,000 audio minutes.
Videofrom $0.000375 per minuteQnHD (320×180). SD is $0.00075 and HD $0.0015.
AI voice agent+$0.002 per minuteYour AI providers bill you directly, with no markup.
Recording+$0.002 per audio minuteSD video recording is $0.006. Off unless you turn it on.
Provider keysBring your ownNo MediaSFU markup on supported AI and SIP providers.
Next best step

See it working before you commit.

Try a demo, open a studio, or start from the apps on your device. The right path should get you to something working fast.