ElevenLabs announced the launch of ElevenAI and the addition of Routines to its ElevenAgents platform during its New York Summit on November 11, 2026. These updates expand the company’s enterprise capabilities from voice technology into meeting assistance and automated operational workflows.
ElevenAI is currently available as a free early-access preview for enterprise teams, offering meeting notes, dictation, and live assistance across 90 languages. Meanwhile, Routines allows agents to execute multi-step operations before, during, and after conversations, such as pre-screening applicants.
ElevenLabs announced ElevenAI and Routines at its New York Summit on November 11, 2026, expanding its enterprise offering from voice technology into meeting assistance and operational workflows. ElevenAI combines meeting notes, dictation, live assistance, summaries, and action items. Routines adds multi-step work around conversations in ElevenAgents. Both appear in the company’s Summit announcement, but their access models differ.
ElevenAI is a preview, with free early access available by request. It is not a generally available app that anyone can immediately download. Routines, meanwhile, is presented as a capability agents can use now inside ElevenAgents, according to the company’s accompanying announcement.
That distinction matters for teams planning a rollout. One announcement offers an entry point for evaluating an employee-facing assistant; the other extends what business agents can do around an interaction.
What Readers Can Access Now
ElevenLabs describes ElevenAI as an app for enterprise teams. Its scope includes capturing information from conversations and helping employees use that information while they work. Routines belongs to ElevenAgents, the company’s platform for configuring and deploying voice and chat agents across functions such as support, sales, recruitment, and operations.
Product
Announced role
Access position
ElevenAI
Meeting notes, dictation, live assistance, summaries, and action items for enterprise teams
Preview; teams can request free early access
Routines in ElevenAgents
Multi-step operations before, during, and after conversations
Described by ElevenLabs as a capability agents can use now
The wording around Routines is more immediate than the language around ElevenAI. In its November 11 LinkedIn announcement, ElevenLabs wrote:
“Agents can now run operations before and after conversations”
That statement supports treating Routines as an added platform capability, but it is not a deployment guide or a promise that every account has identical access. The cited announcements do not explain its rollout, plan eligibility, or configuration process.
For ElevenAI, the available next step is clearer: request early access. TechCrunch’s November 13 report also confirms that users must sign up instead of simply downloading the app.
ElevenAI Brings Meeting Context Into Everyday Work
ElevenAI combines several familiar workplace AI functions in one enterprise-focused product. According to ElevenLabs, it can take meeting notes, produce summaries and action items, support dictation, and provide live assistance.
TechCrunch adds detail about the intended workflow. Its report describes the ability to query information across meetings and retrieve relevant details from earlier conversations during a live discussion. It also reports integration with ElevenAgents for follow-up actions, such as drafting an email from meeting notes.
The useful distinction is between recording a conversation and making its contents available during later work. A transcript provides a record. Cross-meeting queries and live assistance aim to make that record usable without requiring someone to remember which meeting contained a particular detail.
That is a meaningful enterprise proposition. A team can have accurate notes and still lose time finding a decision, constraint, or promised follow-up. ElevenAI’s announced feature set targets that retrieval problem as well as the initial transcription task.
Dictation extends the app beyond meetings. TechCrunch reports support for converting speech into cleaned-up text in any app, across more than 90 languages. It also reports the company’s claim that ElevenAI understands industry-specific terminology in dictation and meeting transcription. The cited coverage provides no independent accuracy measurements, so those capabilities should be treated as announced features, not tested performance.
My view is that the team context is the more consequential part of this product update. Dictation can help one person write faster. Shared meeting context could influence how several people and agents coordinate work. That also makes access controls and information boundaries more important.
Built-In Consent Is a Starting Point
ElevenLabs says meeting notice and consent mechanisms are built into ElevenAI. TechCrunch likewise reports that the app can notify other participants that a conversation is being recorded. These are important details for an assistant designed to capture workplace discussions.
The practical questions extend beyond whether a notification appears. An enterprise evaluation should establish:
How notice and consent are presented to participants.
What happens when someone declines recording.
Who can retrieve a meeting’s contents afterward.
Whether information can pass into an agent or a follow-up action.
These are evaluation questions, not documented ElevenAI settings.
I would treat the built-in mechanisms as a useful starting point, not blanket compliance assurance. Permission to capture a meeting and permission to reuse its contents are separate operational decisions. A participant agreeing to recording does not automatically resolve which colleagues or automated systems should later receive the information.
The official announcement does not publish broad ElevenAI pricing, platform requirements, or a general-availability date. Free early access therefore establishes the cost of the preview offer, not the eventual economics or deployment requirements of a company-wide rollout.
Routines Expands the Work Around a Conversation
Routines addresses a different part of the workflow: the steps surrounding an agent interaction.
ElevenLabs describes multi-step work before, during, and after conversations. Its LinkedIn announcement gives examples including pre-screening a mortgage applicant and calling them back, or sending a text about a confusing bill before the issue becomes a dispute.
Those examples show why this is more than a note-generation feature. A conversation can become one stage in a longer process, with preparation beforehand and follow-up afterward.
The mortgage example links an assessment step with an outbound call. The billing example describes proactive contact about a potential problem. Both suggest a shift from an agent responding within a conversation to an agent participating in an operational sequence.
The examples are vendor descriptions, not published customer results. They do not establish that Routines reduces disputes, improves loan processing, or makes autonomous financial decisions.
The practical opportunity is fewer manual handoffs between conversation handling and follow-up work. The corresponding risk is that an interpretation error can travel further. A mistaken summary is a document error; feeding that mistake into an automated workflow can produce an incorrect message or action.
That is why I would evaluate Routines by the boundaries around its steps, not simply by how many steps it can automate. A useful workflow needs clarity about what information each step receives, what it may change, and where a person must approve an action.
The Bigger Shift Is From Capturing Speech to Acting on It
ElevenAI and Routines approach enterprise work from opposite sides. ElevenAI starts with employees: their meetings, spoken input, questions, and follow-ups. Routines starts with agents and the operations surrounding their conversations.
Together, they show ElevenLabs moving further into the layer between information capture and execution. TechCrunch’s report of ElevenAI-to-ElevenAgents integration supports that direction, although the published material does not provide a complete technical account of how the two products exchange context or permissions.
The strategic expansion is substantial because these products address different business decisions. Choosing a speech model involves recognition quality, latency, and integration. Choosing a meeting assistant also involves employee access and recording policy. Automating follow-ups adds authority to act on business systems or communicate with other people.
That reasoning suggests a practical evaluation sequence:
Check capture quality. Confirm that important names, terms, decisions, and commitments survive transcription and summarization.
Check retrieval boundaries. Establish who can ask about a meeting and what information they can receive.
Separate drafts from external actions. Drafting an email creates a review opportunity; sending it creates an immediate consequence.
Plan for failed or repeated steps. Determine how interrupted workflows are surfaced and how duplicate follow-ups are prevented.
These are recommended checks, not claims that ElevenLabs has documented each control.
For teams exploring these AI products, a bounded trial would be more informative than a broad automation rollout. Start with a workflow whose output a person can inspect, then assess whether the captured context remains accurate as it moves into follow-up work. The valuable test is whether the whole chain stays reliable, from what someone said to what the system is allowed to do with it.
Frequently Asked Questions
3 questions
1
Can Anyone Download ElevenAI Now?
No. ElevenAI is in preview, and enterprise teams must request early access rather than download a generally available app.
2
What Is Routines in ElevenAgents?
Routines adds multi-step operations around conversations, including preparation and follow-up work. ElevenLabs gives examples such as pre-screening an applicant and arranging a callback.
3
Does ElevenAI Include Recording Consent Features?
ElevenLabs says meeting notice and consent mechanisms are built in. Teams should still evaluate how those mechanisms fit their recording, access, and information-sharing policies.
