← All organizations

Realtime audio, video, and voice AI infrastructure

Daily

Daily builds infrastructure for developers to add realtime audio, video, and AI conversations to applications. Its Video SDKs support web, mobile, desktop, and server applications, with recording, live streaming, and transcription capabilities. Pipecat, its open-source framework, lets engineers build voice and multimodal agents using interchangeable models, data stores, and network transports. Pipecat Cloud provides managed deployment and scaling for these agents, supporting uses such as patient intake, scheduling reminders, and automated interviews; the same agent code can also be self-hosted.

Daily was co-founded by Doug Brunton, Nina Kuruvilla, and Kwindla Hultman Kramer, its CEO. Its WebRTC SDKs share a Rust core, allowing updates across platforms through a common implementation. Pipecat grew out of Daily’s internal tooling for conversational AI. Another engineering contribution, Smart Turn, detects when someone has finished speaking directly from raw audio, helping voice agents time their responses. Daily makes the model’s weights, datasets, and training code public.

Pipecat Cloud became generally available in 2026 after a nine-month beta in which Daily reported more than 1,000 teams building and scaling voice agents. Daily’s historical financing includes a $40 million Series B led by Renegade Partners in 2021, bringing funding across three rounds over the preceding 18 months to $60 million.

www.daily.co

Start here

  1. Full Workshop: Realtime Voice AI — Mark Backman, Daily

    Start with Mark Backman and Aleix's workshop to learn the setup and composition of a Pipecat agent using Gemini Live, including credentials, transports, and audio processing.

    Mark Backman · AleixAI Engineer World's Fair 2025

  2. Pipecat Cloud: Enterprise Voice Agents Built On Open Source

    Use this session to understand the operational concerns of enterprise voice agents, from noise cancellation and telephony integration to evaluation and observability.

    Kwindla Hultman KramerAI Engineer World's Fair 2025

  3. The New Primitives: Building AI-Native Software

    For a broader design perspective, learn how computing history and a Knowledge Navigator-style demonstration inform Kwindla Hultman Kramer's argument about software beyond today's agents.

    Kwindla Kramer · Kwindla Hultman KramerAI Engineer World's Fair 2026

  4. Realtime Conversational Video with Pipecat and Tavus — Chad Bailey and Brian Johnson, Daily & Tavus

    Chad Bailey of Daily and Brian Johnson of Tavus explain how to build real-time conversational video agents by combining voice-model pipelines, Pipecat orchestration, and Tavus digital-human video.

    Chad Bailey · Brian JohnsonAI Engineer World's Fair 2025

Messages from the stage

Latency spans the whole conversation

Kwindla Hultman Kramer's voice-bot talk connects transport, transcription, phrase endpointing, inference placement, and synthesis. His joint session with Sean DuBois discusses networking choices through end-to-end voice response time.

From audio pipelines to deployed agents

The workshop introduces interchangeable models and semantic turn-taking. The Pipecat Cloud session extends the engineering discussion to interruptions, asynchronous tools, telephony, evaluation, and observability.

Voice interfaces beyond spoken answers

The joint Daily–Tavus session discusses audiovisual synchronization for conversational video agents. In The New Primitives: Building AI-Native Software, Kwindla Hultman Kramer considers models, data, tools, and context as building blocks for richer multimodal software.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28