← All organizations

AI audio models, creative tools, and conversational agents

ElevenLabs

ElevenLabs builds AI audio models and tools for businesses, creators, and developers. ElevenCreative lets creators and marketers generate, edit, and localize speech, music, images, and video across more than 70 languages. ElevenAgents lets businesses deploy voice and chat agents with integrations, testing, and monitoring for customer operations. ElevenAPI exposes its audio models to developers building interactive products. The broader toolkit includes voice cloning, dubbing, transcription, and sound effects.

Founded in 2022, ElevenLabs is led by co-founders Mati Staniszewski, CEO, and Piotr Dąbkowski, CTO. The two grew up together in Poland, where poorly dubbed films inspired their interest in overcoming language barriers. Its voice research addresses both speaker identity and expressive delivery: voice cloning encodes the target speaker’s characteristics, while voice conversion preserves source intonation. Its multilingual dubbing research applies these ideas to carrying a speaker’s voice, emotion, and delivery across languages.

Customers include Deutsche Telekom, Square, and Revolut for enterprise agents, and Duolingo, NVIDIA, and TIME for creative content. The company reported surpassing $500 million in annual recurring revenue during the first four months of 2026. In February 2026, it announced a $500 million Series D at an $11 billion valuation; the round reportedly exceeded $550 million after its third close.

elevenlabs.io

Start here

  1. Give Your Chat Agent a Voice

    Start with Luke Harries’s talk to learn how Voice Engine preserves existing orchestration, retrieval, and tools, and how client-side and server-side tool calling fit into a speech interface.

    Luke HarriesAI Engineer Europe 2026

  2. Reverse engineering a Viking VoIP phone protocol with Claude Code — Boris Starkov, ElevenLabs

    Read about Boris Starkov’s use of Claude Code to understand how an undocumented device whose configuration software required Windows XP became a working conversational phone demonstration.

    Boris StarkovAI Engineer Europe 2026

  3. Training an LLM from Scratch, Locally

    Use Angelos Perivolaropoulos’s workshop as a starting point for implementing a language model locally, with a nanoGPT-inspired approach, generated outputs, and a participant challenge.

    Angelos PerivolaropoulosAI Engineer Europe 2026

  4. Building Conversational AI Agents - Thor Schaeff, ElevenLabs

    ElevenLabs developer-experience engineer Thor Schaeff leads a hands-on workshop on multilingual conversational AI agents, demonstrating regional accents, language-specific text-to-speech voices, and the Voice Library.

    Thor Schaeff · PaulAI Engineer World's Fair 2025

Messages from the stage

Designing spoken interaction

Schaeff’s workshop addresses regional accents, specialized pronunciation, latency, and multilingual transcription limits. Harries adds emotion-aware turn-taking and semantic VAD, while Reeve’s statue demonstration raises questions about culturally appropriate accents and conversational interruption.

Testing an undocumented phone protocol

Starkov describes brute-forcing device commands and investigating settings that vanished after reboot. His account emphasizes validating discoveries through closed-loop experimentation before routing the phone through a Twilio SIP trunk.

Understanding training behavior

Perivolaropoulos’s workshop connects tokenizer and transformer implementation with optimization schedules, validation, overfitting, and initial loss behavior in a small GPT-2-style model built using Torch.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-27