ElevenLabs launches custom AI tuning, ElevenAI app, and marketplace with Nvidia expansion
ElevenLabs has launched a series of enterprise-focused AI tools, including custom model fine-tuning, a workplace app called ElevenAI, and an expanded partnership with Nvidia. The company also introduced a partner programme and marketplace, alongside new speech-to-text models, Scribe v3 and Scribe v3 Live, aimed at improving accuracy and real-time performance.
Editor, Lazyfounder
TABLE OF CONTENTS
- ElevenLabs expands enterprise AI tools with custom tuning and workplace app
- Custom AI tuning and ElevenAI app target enterprise productivity
- Partner programme, marketplace, and Nvidia deal strengthen enterprise offerings
- Scribe v3 and Scribe v3 Live improve speech-to-text accuracy and real-time performance
- Accuracy comes at a cost: Scribe v3 lags in speed and price
- Sources
ElevenLabs expands enterprise AI tools with custom tuning and workplace app
ElevenLabs unveiled a suite of enterprise-focused AI initiatives at its New York summit, targeting conversational agents and workplace productivity. The announcements include custom AI model fine-tuning for businesses using their own customer conversation data, the launch of ElevenAI—a workplace app that converts meetings into notes and to-do lists—and a partner programme and marketplace for third-party applications. The company also expanded its partnership with Nvidia, adding 15 megawatts of computing power to support large-scale enterprise deployments.
Custom AI tuning and ElevenAI app target enterprise productivity
The custom AI model fine-tuning allows businesses to improve ElevenLabs’ voice models using their own customer interaction data. Companies can collaborate with ElevenLabs researchers to set specific goals, such as reducing errors or increasing customer satisfaction. The models adapt and improve as more data becomes available, enabling enterprises to deploy conversational agents tailored to their needs.
ElevenAI, the newly launched workplace app, automates meeting documentation by transcribing discussions and generating notes and action items. The app requires explicit consent from all participants before recording, addressing privacy concerns in professional settings. Free early access to ElevenAI is now available.
Partner programme, marketplace, and Nvidia deal strengthen enterprise offerings
ElevenLabs introduced a partner programme and marketplace, featuring over 550 partners across 44 countries, including Google Cloud and Deloitte. The marketplace, set to open next year, will allow businesses to purchase ready-made applications built by partners, such as Salesforce, IBM, and Zendesk. ElevenLabs also announced a partnership with McKinsey, further expanding its enterprise reach.
The company’s collaboration with Nvidia deepened, with an additional 15 megawatts of computing power allocated for large customers. This expansion supports the development of a voice agent for Poland’s national health booking system, which will contact patients using landlines. ElevenLabs is also integrating Nvidia’s open Nemotron 3.5 Super VL language model into its ElevenAgents platform, enabling advanced use cases like local voice control for robots. Jensen Huang, Nvidia’s CEO, was scheduled to appear at the summit but withdrew due to an ongoing board inquiry into his conduct.
Scribe v3 and Scribe v3 Live improve speech-to-text accuracy and real-time performance
ElevenLabs released Scribe v3 and Scribe v3 Live, two new speech-to-text models designed for accuracy in challenging conditions, such as meetings with background noise or overlapping speakers. Scribe v3 is optimised for pre-recorded audio, while Scribe v3 Live targets real-time applications with low latency.
According to ElevenLabs, Scribe v3 reduces word errors by 25% across 102 languages compared to its predecessor, with significant improvements in German accuracy and speaker diarization—particularly in group conversations. Scribe v3 Live introduces real-time speaker diarization and analyzes conversational context to avoid cutting off callers mid-sentence, a common issue in voice agents.
Accuracy comes at a cost: Scribe v3 lags in speed and price
Independent benchmarks confirm Scribe v3 as the most accurate transcription model available, trailing only Alibaba’s preview model. However, Scribe v3 is slower and more expensive than competitors like Microsoft and xAI, which offer faster processing at marginally higher error rates and roughly half the cost. The benchmark price for Scribe v3 is $3.33 per 1,000 minutes.
Both models offer customisation options, such as preloading domain-specific terminology and formatting rules for numbers or dates. Enterprise customers can opt for isolated servers in the EU and a zero-retention mode, which deletes data immediately after processing.
Related on Lazyfounder
Sources
- The Next Web · 2026-11-11
ElevenLabs moves beyond voice with custom models and a work app - The Decoder · 2026-11-12
ElevenLabs Scribe v3 Live uses conversational context to stop voice agents from cutting off callers mid-sentence
This story is an original summary drafted with AI by Lazyfounder from the reporting listed above and checked by automated validation. Facts are attributed to their original publishers; sections marked as analysis are Lazyfounder's. Where a source is in another language, facts were machine-translated and quotations are reported, not reproduced. Read the original coverage via the links, and see our AI policy and corrections policy.
About the author
Editor, Lazyfounder
Tarun Mottlia edits LazyFounders, covering Indian startups, funding rounds, AI and product launches. Every story on the site is AI-assisted and checked against its cited sources before publication.
More stories by Tarun MottliaGet the LazyFounder Brief
Startup, funding and AI news in a five-minute read. Join the early-access list.
