ElevenLabs pushes voice as the default AI interface as wearables go always-on

Voice is quickly moving from a “nice-to-have” feature to the control layer for AI products, especially as assistants spread into headphones, cars, and smart glasses.

Voice models pair with LLM reasoning and agentic workflows

ElevenLabs CEO Mati Staniszewski argues the big shift is not just more realistic speech, but voice systems working with LLMs (large language models) so they can reason, remember context, and handle multi-step tasks. For marketers and e-commerce operators, that points to more hands-free workflows: asking an assistant to pull campaign performance, summarize customer feedback, or draft variants while you are away from a keyboard.

Iconiq Capital partner Seth Pierrepont made a similar point at Web Summit in Doha, saying traditional inputs like keyboards are starting to feel dated, while screens will still matter for entertainment and gaming. The practical implication: UX design for business software may need to support “conversational control” alongside dashboards.

Staniszewski also highlighted a more agentic direction: instead of explicit prompting every time, systems can rely on guardrails, integrations, and persistent context accumulated over time. In practice, that means voice assistants that already know your brand voice, preferred reporting cadence, and which tools they can safely access.

Hybrid cloud plus on-device audio raises privacy stakes

Historically, high-quality speech generation and understanding have mostly run in the cloud. ElevenLabs says it is working toward a hybrid approach that blends cloud with on-device processing, aimed at new hardware like headphones and other wearables where voice becomes a constant companion.

The company is already partnering with Meta to bring its voice tech into products including Instagram and Horizon Worlds, and it signaled openness to future work on Ray-Ban smart glasses as voice-first interfaces expand.

The tradeoff is obvious: always-on voice increases privacy and surveillance concerns, especially as companies store more personal context. Those risks are not theoretical—Google has faced accusations that its voice assistant spied on users, leading to a settlement reported here: https://techcrunch.com/2026/01/26/google-pays-68-million-to-settle-claims-its-voice-assistant-spied-on-users/.