Webinar
Originally aired
Master Call Flows with SignalWire + WebRTC
Simplify your real-time communication apps using WebRTC and SignalWire together.
Subscribe
Transcript
00:00 | So yeah, I'm going to be showcasing two technical demos today utilizing WebRTC and some of the other tools that we discussed such as SWML scripts and subscribers. Um the first demo will be what's called Lexi call. It is essentially a AI agent that will act as a all- knowen librarian who will have the skills to be able to look up your query via Wikipedia or just a general search engine uh query. Uh the second demo I will be demonstrating is if you were here for the last live wire, you might be familiar with the live context |
00:33 | transfer. I'll be kind of going over that demo again, but uh also showcasing a special sauce I've added to it for uh controlling your WebRTC call via Swiml scripts. So let me go ahead and share my screen and we could get started. So here's Lexi call. Um, it's a builtout librarian website that we have going on. It has a click to call button right here. So, when we want to actually talk to Lexi, all we have to do is just click this button. So, we're going to go ahead and do that. |
01:14 | There we go. Hello. It seems like you're ready to dive into something. What topic are you interested in exploring today? Yeah, I'm I'm interested in exploring free switch. Can you tell me more about free switch? Looking that up in the encyclopedia. Free switch is a free and opensource telefan software designed for realtime communication protocols. It supports various media forms including audio, video, and text. Some of its key applications include web RTC, voice over internet protocol, VoIP, |
01:57 | video transcoding, and multiple. I would actually like to learn more about Signal Wire now, but they're not on the Wikipedia. Can you do a web search for that? Signal Wire is a platform that provides APIs for voice, messaging, and video communications. It enables developers to create ultra low latency, high reliability communication applications. Some of the key features include one programmable unified communications. Signal wires platform allows B and then also with WebRTC since we can tap into the events of the call, we can |
02:41 | actually immediately redirect you or pop up a menu uh upon that event firing over the SDK. So here I had it actually uh navigate us to a survey that we could take we could take um about our experience talking with the AI agent. So that's our first demo. Um, our second demo is going to be last live wires with the secret sauce that I've added. Um, for again context, this is going to be contextaware call transfers. Essentially, the flow is going to be I am going to call into a AI agent from a |
Key takeaways
WebRTC turns any browser into a voice endpoint, so callers reach your AI agents with one click
Persistent AI agents can gather and verify caller information before a human ever picks up
Passing conversation context to live agents eliminates blind transfers and repeated questions
About this webinar
In this excerpt from our LIVEWire webinar, Mastering Call Flows with SignalWire + WebRTC, we walk through two live demos that show how to build rich, real-time voice experiences using SignalWire's AI agents, WebRTC, and SWML (SignalWire Markup Language), a declarative markup for call control.
The first demo introduces LexiCall, an AI librarian that answers spoken questions using Wikipedia and web search, like a digital reference desk powered by voice. A click-to-call button on a webpage connects users directly to Lexi, who responds conversationally with real-time summaries. Because it runs on WebRTC, everything happens in the browser with no downloads or plugins.
The second demo shows context that travels with the caller. An AI receptionist greets the caller, checks membership status, walks new users through signing up, and collects a summary of their issue. The call then transfers to a live agent who receives the full conversation context before picking up, making handoffs smoother and support smarter.
What you'll learn
How to embed click-to-call voice AI in any webpage using WebRTC, with no downloads or plugins
How to build an AI agent that answers spoken questions with live data from Wikipedia and web search
How to design call flows where customer context transfers with the call to a live agent
How to script greetings, membership checks, data collection, and live call transfers with declarative markup
Speakers
Who is this webinar for?
Developers and technical teams building voice applications, from browser-based calling to AI-assisted support flows. Familiarity with web development helps, but the demos explain each step.
What is SWML?
SWML (SignalWire Markup Language) is declarative markup for controlling calls. It scripts behaviors like greetings, data collection, and transfers without managing low-level telephony.
Do I have to learn SWML to build AI voice applications?
No. The Agents SDK lets you control call behavior with Python instead of writing SWML directly.
Do callers need to install anything?
No. WebRTC runs natively in modern browsers, so a click-to-call button is all users need.
Can live agents see real-time transcription during a call?
Yes. The live_transcribe verb makes real-time transcription available to human agents.
Can I customize the AI agent's voice?
Yes. SignalWire supports multiple text-to-speech providers, including Rime and ElevenLabs.
Related Resources
