ElevenLabs has unveiled 13 new products and upgrades at its Summit, making its biggest push yet to turn voice AI from a novelty into everyday enterprise infrastructure. The package spans three new voice models, a system for orchestrating them, and a new app that acts as an AI teammate for office teams.
“Within a few years, conversational agents will become critical infrastructure for every enterprise, as essential as cloud or payments,” said Mati Staniszewski, ElevenLabs’ co-founder and chief executive. “Our goal is to build the interaction layer for AI.”
Three new voice models lead the launch
Over the past month, ElevenLabs has shipped three new voice models, all of which the company says lead independent benchmarks. Scribe v3 is a speech-to-text model that transcribes what people say, Eleven v4 is a text-to-speech model that reads tone, pacing and context from a script, and Eleven Realtime is a new class of speech-to-speech models built for fast, natural voice agents.
The company is also pushing into industry-specific models. It launched Scribe v3 Medical, a transcription model tuned for healthcare, and said it plans similar models for sectors such as financial services. Select enterprise customers will also be able to work with ElevenLabs’ research team to fine-tune audio models and large language models on their own data, using live deployments and reinforcement learning to improve measures such as customer satisfaction.
Interaction Engine picks the right model for each step
Alongside the models, ElevenLabs unveiled Interaction Engine, a system that orchestrates voice models and large language models across a conversation. The idea is that not every step of an agentic task needs a frontier model: the engine brings in the best model for each use case, balancing performance and reliability against cost.
“Interaction Engine orchestrates all this under the hood, so you can be confident you’re delivering the right outcome for the customer, without wasting money on tokens,” Staniszewski said.
ElevenAgents, the company’s platform for voice and chat agents, also gets three upgrades. Memory gives agents the full context of every returning customer, recalling details from a new shirt size to a pending move so nobody has to repeat themselves. Routines, a new automation layer, runs multi-step work before, during and after every conversation. Architect is a built-in expert that helps teams build and improve agents by talking or typing, with every change drafted, tested in simulation and held for approval.
A new app, a partner push and a bigger Nvidia bet
The company also opened early access to ElevenAI, a new app that acts as an AI teammate for enterprise teams. It can turn spoken conversations, including in-person meetings, into notes and next steps, with notice and consent mechanisms built in, and offer live assistance during a call.
To speed up deployment in large companies, ElevenLabs launched a Partner Program with more than 550 partners in 44 countries, announced an alliance with McKinsey, and expanded its partnership with Nvidia. Nvidia’s Nemotron 3.5 Super VL model is now available in ElevenAgents, ElevenLabs will add 15 megawatts of new Vera Rubin compute capacity, and the two companies will work on an on-premises voice agent for Poland’s national e-Health platform.
A new Marketplace will let companies adopt industry-specific agents built by more than 20 ElevenLabs partners instead of building from scratch. At the Summit, enterprises showed off voice agents already in use, including Yahoo’s voice-powered AI assistant and Westgate’s new front-desk agent, joining Stripe, TELUS, FOX Sports, Klarna and SevenRooms by DoorDash.



