ElevenLabs previews ElevenAI and Scribe v3 for enterprise teams

Daniel Okafor

ElevenLabs put ElevenAI into preview at its 11/11 Summit in New York. The experimental app combines three functions: meeting notes, dictation and agent follow-up. Access requires early-access registration. It does not yet provide a direct download.

ElevenLabs disclosed no general-availability date, price, customer list or deployment scale, leaving the commercial scope of this week’s launch undefined for enterprise buyers evaluating a broader platform purchase.

What does ElevenAI do?

• Captures meeting notes across multiple meetings.
• Notifies other participants when recording begins.
• Queries data across previous meetings.
• Recalls earlier details through live chat assistance.
• Integrates ElevenAgent for follow-up actions.
• Writes draft emails from meeting notes.
• Converts dictation in any app across more than 90 languages.

ElevenLabs said the system can “understand industry-specific terms across dictation and meeting transcription,” but the claim arrives without accuracy rates, latency measurements, benchmark comparisons or independent validation.

The bundle covers functions now common across the category. The announcement specifies only participant notification for recording. It gives no retention rules, access controls or consent workflow.

What does team-first positioning mean?\n

ElevenLabs carries a $22 billion valuation. It describes ElevenAI as “a product that lives within a team.” The product targets team use, not individual productivity. The valuation does not establish product revenue, adoption or savings.

Fireflies, Wispr Flow and Granola shipped competing capabilities over the last few months, yet ElevenAI remains gated behind early access and lacks disclosed customer evidence. ElevenLabs is the newest entrant in that category.

ElevenLabs presents a license-substitution case. Enterprises could use ElevenAI instead of buying separate notetaker licenses. No seat price or bundle terms emerged. No total-cost comparison appears. The license-avoidance case remains uncalculated.

Meeting tools are moving toward centralized company knowledge hubs. They now aim to automate knowledge extraction and follow-up actions. Governance becomes a buyer question as a result. The launch reports no ElevenAI governance controls.

How does ElevenAgent change the workflow?

ElevenAgent forms the bridge from a meeting record to follow-up action, although the launch cites only a draft email and leaves supported actions, approval rules, permission boundaries and error logging unspecified.

ElevenLabs says its agent platform improves context funneling. It does not identify the orchestration architecture. Enterprise-system support is also unspecified.

Cross-meeting queries imply retrieval over prior records. The announcement gives no retrieval method or source-indexing policy. Accuracy testing is likewise absent.

What did Scribe v3 add?

Last week, ElevenLabs launched the v3 series of Scribe speech-to-text models.

• Adds live transcription to Scribe v3.
• Fine-tunes models on company data.
• Selects a language model by use case.
• Automates transcript-based tasks through a routine creator.

Each component moves Scribe beyond live captions toward controlled enterprise workflows, but no supported fine-tuning volume, model-selection criteria or routine-automation accuracy figure was disclosed in the announcement.

Fine-tuning could improve fit to internal vocabulary. ElevenLabs provides no fine-tuning datasets or deployment tiers. It lists no Scribe-specific language coverage. The 90-plus-language figure applies to ElevenAI dictation, not Scribe v3.

ElevenLabs does not say whether Scribe v3 powers ElevenAI. Buyers cannot treat the launches as one confirmed package.

What remains unproven?\n

Together, the two announcements connect meeting capture, retrieval, agent action, customization and routine automation, but they disclose no pricing, customer count, deployment metric or cost saving.

ElevenAI remains in preview. Scribe v3 has no disclosed customer results. Enterprise buyers have no basis to calculate savings. For them, the decision depends on a missing commercial file: price, deployment count, retention terms and measured accuracy. The launch provides none of those.