The WebRTC.ventures Blog.

Building a telehealth platform starts with a video call, scheduling, and a patient portal, and those follow established engineering patterns. The harder engineering comes from healthcare regulation. Each rule that governs a remote visit becomes a requirement the software has to meet. The platform has to know which state the patient is in during the visit, which clinicians are licensed to see them there, whether a prescription needs extra identity checks, and which vendors are

In the ongoing debate around WebRTC vs. MoQ, the most practical answer may be, “why not both?” The IETF is now examining whether QUIC offers real-time media applications any advantage over WebRTC, with WebTransport and MoQ positioned as possible foundations for the next generation of real-time protocols.  In this episode of WebRTC Live, we welcomed back Lorenzo Miniero, Chairman at Meetecho and author of the Janus WebRTC Server. Lorenzo has been experimenting with MoQ and

A real-time AI avatar gives a voice agent a face, and it also puts a third-party media service between your text-to-speech and the user. In a Pipecat pipeline, the avatar service receives your synthesized speech, renders the face, and sends back the audio the user hears. When the user interrupts, Pipecat tells that service to stop, but speech already in flight still plays. We tested four avatar APIs in avatar-only mode, all through their Pipecat

Deepgram has announced that it is acquiring Daily, the WebRTC infrastructure provider and a long-time WebRTC.ventures partner. We see this as a positive step for developers building real-time voice AI. Deepgram is an enterprise voice AI platform and one of the providers our team uses for speech-to-text (STT) and, more recently, text-to-speech (TTS). The deal brings together two companies we know and value. What the Deepgram acquisition of Daily (and Pipecat) includes Deepgram announced the

WebRTC was built for browsers, but many WebRTC connections now run through server programs, for example SFUs, gateways, cloud gaming, recording, or AI agents that join calls. Without a browser to handle the connection, each of these programs needs its own WebRTC implementation, usually a library built into the service. Google’s libwebrtc, the most widely used implementation, was built for browsers and native client apps. Its API centers on capturing, encoding, and rendering media for

When Jeff Pulver invited me to attend Voice and Conversations on the Net (VON) 2026 on behalf of WebRTC.ventures, I accepted immediately. Three days with people building the next generation of voice communications was an easy yes. While at the conference earlier this week, the enthusiasm for voice AI was clear, and it extended well beyond the companies building the agents. Telecom providers, infrastructure engineers, startup founders, and enterprise teams were all part of the

Microsoft will retire Azure Communication Services (ACS) as a standalone offering on September 30, 2028. After that date, Microsoft supports ACS video only in applications that connect to Microsoft Teams. Companies that use ACS to embed video in their own products will need a new platform, and they have just under two years to move. Microsoft announced the retirement on September 23, 2026, in a Microsoft Learn retirement and breaking changes guide. It describes the

An AI chatbot built on retrieval-augmented generation (RAG), where it answers by pulling from your own knowledge sources, is only as safe as the content it is allowed to retrieve. The failure happens when retrieval runs with no notion of who is asking: the agent searches the knowledge base, ranks by relevance, and returns the best match to any authenticated user. The moment that knowledge base holds mixed-sensitivity content, such as HR salary records next

Before switching video SDKs, upgrading to a new version or renewing a vendor contract, engineering leaders want to know how their call quality compares with competing products and where it falls short. WebRTC benchmarking answers that by running each video calling product through identical network limits, such as a 500 kbps cap, and recording call quality on the receiving device. A benchmark can look correct and be wrong. We ran one on Google Meet and

As Voice AI becomes more prevalent, applications need a way to store conversation data that supports observability, auditing, and long-term retention. The audio recording or a text transcript alone leaves out critical context. Additional details matter too, such as who was in the conversation, what devices were used, what was said, and how decisions were made. vCons act as the “memory” of a conversation. They are structured, portable, machine-readable objects that capture conversation data across

For the third year in a row, I had the pleasure of attending the RTC.ON conference where our company WebRTC.ventures was a sponsor. As usual, the panel was stacked with great talks and speakers. The topics ranged from AI, computer vision, voice, audio, telephony, broadcast and transport protocols, including the ever more trending MOQ, which was the most prominent topic at the conference. You can find the agenda here and the live streams here: day

WebRTC and SIP both carry live voice to an AI agent. Voice AI platforms list them side by side, so teams building an AI receptionist often expect to pick one or the other for connecting callers to the agent. In practice, each protocol serves a different way customers reach the business. An AI receptionist usually has two front doors. The first is your business phone number: a customer calls from a mobile phone or landline,

If you run WebRTC at scale, a large share of your media goes through TURN relays. Restrictive NATs, enterprise firewalls, and clients that hide their IP addresses all push sessions onto relayed candidates, and when TURN server performance drops under load, call quality drops with it. This post shows how eBPF TURN offload moves ChannelData relay into the Linux kernel so media packets never reach the userspace process. We demonstrate it with a TURN server

On September 22–23, I was in Chicago for the IIT Real-Time Communications Conference at Illinois Tech, an annual IEEE conference where industry engineers and academic researchers present work on real-time communications. This year's tracks covered programmable networks, WebRTC applications, emergency communications, and AI in real-time communications. I had the privilege of presenting a talk in the WebRTC track titled "eBPF Makes Python a Good Language for a TURN Server." Across the other sessions and tracks,

A production voice agent has a latency budget of roughly 800 milliseconds before a conversation starts to feel slow. That budget is spent across eight sequential stages, typically split across three to five different vendors depending on how the stack is assembled. This post maps that voice AI pipeline stage by stage: mic capture, WebRTC transport in, turn detection, speech-to-text, LLM inference, text-to-speech, WebRTC transport back, and playout. Each adds its own piece to the

LiveKit can run on Kubernetes without hostNetwork, which lets self-hosted LiveKit scale like any other workload and run in clusters that don't allow host networking. Its pods stay private behind an ordinary ClusterIP Service, and STUNner, a Kubernetes-native TURN server, provides the single public address that WebRTC media flows through. LiveKit is open source and distributed by design, which makes it a natural fit for Kubernetes. LiveKit's Kubernetes guide, though, deploys the server with host

Once again, WebRTC Live broadcasted direct from the RTC.ON conference in Kraków, Poland where WebRTC.ventures CTO Alberto Gonzalez interviewed two speakers: Luke Curley,  MoQ co-creator at moq.dev, ex Discord, ex Twitch. Luke’s talk, “On a Boat: Why Pull-Based Media Streaming Beats Push,” argues that pull-based protocols like MoQ handle bandwidth-constrained, high-camera-count streaming better than push-based options like WebRTC, SRT, and RTMP, covering multipath QUIC, AI-driven rendition switching, and lossy live plus pristine VOD from a

The most frustrating thing about speaking with a Voice AI agent is when it interrupts you, or it doesn’t understand the normal flow of human conversation. There are multiple ways to solve this depending on your use case, and coming up with the best architecture for your needs is crucial to creating positive customer experiences. For over a decade, our team has built voice and video applications for our clients using WebRTC. This has involved

Self-hosting a voice AI pipeline gives you control that managed APIs can't offer: you choose the models, you decide where they run relative to each other, and you keep audio data within your own infrastructure. Getting there means managing GPU workloads, model placement, and service networking yourself. Kubernetes provides the underlying mechanisms for GPU scheduling, autoscaling, and service discovery. Amazon EKS packages them into a managed platform, using Karpenter for GPU provisioning, so you get

Now that everyone's back from summer break here in Madrid, it feels like the right time to recap AI Tinkerers Madrid: AI in Real-Time Communication Systems, which WebRTC.ventures sponsored and hosted this past July alongside BTS, who provided their offices and a great spread of local snacks. We aimed for 30 attendees and comfortably surpassed that number. The feedback was strong, too! People are already asking about the next edition, and I'll have news on