ElevenLabs CEO Says AI Disclosure Is Needed Now as Firm Reportedly Valued at $22B

TechCrunch reported that ElevenLabs builds the voice layer of AI, models that turn text into speech that sounds human. Many people encounter it in customer service without realizing it. Klarna runs first-line phone support for 35 million U.S. customers on it, and Deutsche Telekom, Cisco, Adobe and a growing list of governments also use it. ElevenLabs also sells to creators for audiobooks, dubbing and music.

The company is not alone. TechCrunch reported it is increasingly bumping into customers, including Decagon, a conversational AI platform that trained its voice product on ElevenLabs and now competes with it. Investors appear unconcerned: the four-year-old company says it is pacing at $600 million in ARR and is now reportedly valued at $22 billion by its backers.

Staniszewski spoke at Nrth in Toronto, a local entrepreneurship conference formerly known as Elevate. He said audio models are still far from commoditized. He had predicted at TechCrunch Disrupt last year that audio models would be commoditized within a couple of years. Asked to rate that prediction, he said there is still a lot of work to be done and that the quality delta achievable at the model level remains significant. In three to five years, he said, those differences will probably be smaller.

He said ElevenLabs wants to be first to pass the Turing test for conversational AI. That requires combining intelligence with emotional intelligence, he said, including understanding the other side's emotions and knowing when to slow down or speak up. That has not yet been done.

On the business mix, Staniszewski said 55% or more of ElevenLabs's revenue is classic enterprise, while a large percentage of the remaining 45% comes from small and medium businesses, developers, builders and creators.

Asked about competition with customers, he said the lines between model companies, platform companies and application companies are becoming more blurry. In the past, there were clear splits between where one started and another ended. Today, he said, that line is much less clear. He pointed to Anthropic as an example of a model company that is definitely a platform and increasingly a wide set of applications.

TechCrunch reported that ElevenLabs customers can choose a "reasoning layer" from a menu of options. Staniszewski said the choice between frontier lab models and open-weight models is less binary. In customer experience, if a caller only needs information and no actions are executed, many open-source models can be used because the knowledge base defines what a good experience is. But in financial services, where authentication, transaction information and possible refunds are involved, there is no room for error, and frontier models will still lead.

Some open-weight models are Chinese. The U.S. government is a customer, as are European governments, according to the interview. Staniszewski said conversations differ by deployment. The models and voices ElevenLabs deploys depend on the case. If it works with the Polish or Brazilian government, they have their own requirements, which can mean an open-weight model, a closed-source model or their own fine-tuned model. In Poland, he said, the use case is healthcare: patients booking appointments across the public health system, with 18% never showing up. The deployment uses agents that call and remind patients. They had a set of models optimized on their knowledge, and ElevenLabs integrates while keeping data residency.

On disclosure, Staniszewski said there should be disclosure at this time. People are not used to AI agents, and the common pattern is that customers do not want to feel cheated on a call. But in five years, when everyone has their own agent working on their behalf, callers will expect an agent, and society will shift, he said. He described good ways to handle the transition, such as offering customers a choice if there is a 30-minute wait for a human. In almost all cases, he said, they choose the agent and are then surprised by how good the experience is.

Asked about gross margins, Staniszewski said he could not discuss them in detail, according to TechCrunch. He was clear that he does not mind them getting squeezed further if it means expanding the company's market share.

Editor's Summary

ElevenLabs CEO Mati Staniszewski told TechCrunch that businesses should disclose when customers are speaking with AI agents, though he expects that norm to change as personal agents become common. TechCrunch reported the four-year-old voice AI company is pacing at $600 million in ARR and is now valued at $22 billion by its backers, with more than half its revenue from enterprise customers. Staniszewski also said model and application lines are blurring, and that frontier models remain preferred for high-stakes tasks like financial services.