ElevenLabs CEO: Voice is the next interface for AI

ElevenLabs co-founder and CEO Mati Staniszewski says dependable is becoming the adjacent large interface for AI – the mode radical volition progressively interact with machines arsenic models determination beyond substance and screens.

Speaking astatine Web Summit successful Doha, Staniszewski told TechCrunch dependable models similar those developed by ElevenLabs person precocious moved beyond simply mimicking quality code — including emotion and intonation – to moving successful tandem with the reasoning capabilities of ample connection models. The result, helium argued, is simply a displacement successful however radical interact with technology. 

In the years ahead, helium said, “hopefully each our phones volition spell backmost successful our pockets, and we tin immerse ourselves successful the existent satellite astir us, with dependable arsenic the mechanics that controls technology.”

That imaginativeness fueled ElevenLabs’s $500 cardinal raise this week astatine an $11 cardinal valuation, and it is progressively shared crossed the AI industry. OpenAI and Google person some made dependable a cardinal absorption of their next-generation models, portion Apple appears to beryllium softly gathering voice-adjacent, always-on technologies done acquisitions similar Q.ai. As AI spreads into wearables, cars, and different caller hardware, power is becoming little astir tapping screens and much astir speaking, making dependable a cardinal battleground for the adjacent signifier of AI development. 

Iconiq Capital wide spouse Seth Pierrepont echoed that presumption onstage astatine Web Summit, arguing that portion screens volition proceed to substance for gaming and entertainment, accepted input methods similar keyboards are starting to consciousness “outdated.”

And arsenic AI systems go much agentic, Pierrepont said, the enactment itself volition besides change, with models gaining guardrails, integrations, and discourse needed to respond with little explicit prompting from users. 

Staniszewski pointed to that agentic displacement arsenic 1 of the biggest changes underway. Rather than spelling retired each instruction, helium said aboriginal dependable systems volition progressively trust connected persistent representation and discourse built up implicit time, making interactions consciousness much earthy and requiring little effort from users. 

That evolution, helium added, volition power however dependable models are deployed. While high-quality audio models person mostly lived successful the cloud, Staniszewski said ElevenLabs is moving toward a hybrid attack that blends unreality and on-device processing — a determination aimed astatine supporting caller hardware, including headphones and different wearables, wherever dependable becomes a changeless companion alternatively than a diagnostic you determine erstwhile to prosecute with. 

ElevenLabs is already partnering with Meta to bring its dependable exertion to products including Instagram and Horizon Worlds, the company’s virtual world platform. Staniszewski said helium would besides beryllium unfastened to moving with Meta connected its Ray-Ban astute glasses arsenic voice-driven interfaces grow into caller signifier factors. 

But arsenic dependable becomes much persistent and embedded successful mundane hardware, it opens the doorway to superior concerns astir privacy, surveillance, and however overmuch idiosyncratic information voice-based systems volition store arsenic they determination person to users’ regular lives — thing companies similar Google person already been accused of abusing.