Features and Architecture
This section provides an overview of Custom AI Companion's features and general, dataflow diagrams.
Custom Meeting Summary Templates
Custom Meeting Summary Templates: Summaries designed for the discussion at hand
Zoom’s Custom Meeting Summary Templates allow account admins to tailor meeting summaries to fit different needs and audiences. Instead of relying on a standard format, organizations can create templates that highlight the most relevant details for specific use cases.
For example, a Customer Success team could implement a Customer Experience Review Summary template that automatically organizes discussions around key performance metrics including Customer Satisfaction (CSAT) scores, Net Promoter Score (NPS) trends, Customer Retention Rate analysis, and churn risk indicators. This template could structure summaries to capture quarterly performance comparisons, identify emerging satisfaction patterns, highlight successful retention strategies, and document action items for addressing customer experience gaps.
Custom Meeting Summary Templates Architecture
The following diagram outlines typical dataflows for Custom AI Companion’s Custom Meeting Summary Templates under Zoom’s federated approach.
Custom Dictionary
Custom Dictionary: Provide AI Companion with your business’s or industry’s language and jargon for better transcripts
Custom Dictionary enhances AI Companion's ability to recognize and incorporate your organization's specialized terms, industry jargon, and acronyms, resulting in more accurate and contextually relevant meeting transcripts and summaries.
When utilized, Zoom’s live transcription service will reference the account’s Custom Dictionary during the live transcription process, resulting in more accurate meeting transcripts and summaries.
This feature is especially beneficial for industries that rely on specialized language, such as healthcare, finance, or manufacturing sectors, where precision in communication is crucial. A Custom Dictionary is designed to help AI Companion adopt the unique vocabulary of the organization, improving the relevance and quality of AI-generated responses.
For example, imagine a legal team using AI Companion to generate a meeting summary from a strategy session where one attorney says, “We need to prepare a memorandum regarding the Daubert standard and its applicability in this tort case.” Without a legal custom dictionary in place, it can be difficult to accurately identify uncommon words like “Daubert,” potentially rendering it as “Dahlberg” or “dog bird,” which could lead to confusion or misinterpretation. However, with a domain-specific dictionary that defines terms like “Daubert,” “tort,” and “voir dire,” AI Companion is better equipped to recognize and transcribe specialized language.
Custom Dictionary supports up to 500 unique words in English
Organizations can currently incorporate up to 500 specialized terms into their Custom Dictionary for English-language meetings.
Custom Dictionary Architecture
The following diagram outlines a typical dataflow for a Custom Dictionary applied to a meeting’s transcript to generate a meeting summary under Zoom’s federated approach.
Custom Avatars
Custom Avatars: Bring your scripts to life with dynamic AI avatars
Custom AI Companion’s Custom Avatars feature enables users to create a personalized virtual avatar using their recorded video and voice, or select from a list of pre-provided avatars. Once an avatar is selected or generated, users can upload a script that is transformed into a Zoom Clip, with the avatar narrating the content in a natural voice and, if applicable, the user’s likeness.
This feature is especially useful for businesses looking to scale video production efficiently, allowing users to create training materials, lessons, and presentations with just a transcript—reducing production time while maintaining quality, personalized video content.
For example, a compliance officer may need to remind employees of quarterly training requirements. Rather than recording a new message or sending another lengthy email, they use their custom avatar to generate a brief, direct video reminder from a script—quickly reinforcing the message while keeping communication efficient and repeatable.
Custom avatars also support multiple languages, enabling users to upload a script in any supported language and generate output in the same language, or produce a copy in an alternative language. For example, if a user records their custom avatar likeness in English, they can upload a script in Spanish or Russian and will receive a clip spoken in the corresponding language that reflects their authentic voice. Alternatively, a user can upload a script in English, and also receive copies of the clip in Spanish or Russian. The current list of supported languages includes:
Chinese (Simplified)
Chinese (Traditional)
Dutch
English
French
German
Indonesian
Italian
Japanese
Korean
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Users with a Custom AI Companion license are allotted six minutes of custom clips a month and refreshes on their account’s monthly billing date
Users with a Custom AI Companion license are allotted up to six minutes of Custom Avatar Clip generation each month. This allocation resets on each account’s monthly billing date, so if an account’s billing date falls on the 15th, users who generate their allotment of clips on or before the 14th will receive a refreshed allotment the following date.
Additional Custom Avatar Clip time can be purchased with an add-on
Customers interested in granting certain users more than six minutes of clip generation time per month may purchase an add-on for additional clip generation time. For more information on this add-on, speak with your Zoom account team.
Custom Avatar Architecture
The following diagrams outline typical dataflows for using Custom AI Companion’s Custom Avatars feature under Zoom’s federated model.
Avatar Creation
The following diagram outlines the process for creating a custom avatar without generating a clip.

Clip Creation
The following diagram outlines the process for creating a clip using a pre-existing avatar.

Knowledge Collections
Knowledge Collections: Enable AI Companion to access and reference organizational content
Knowledge Collections empower organizations to connect company documents, resources, and/or existing search infrastructure (i.e., a third-party solution with indexing capabilities) into Zoom AI Companion, enabling it to generate more contextually relevant responses. Custom AI Companion allows companies to connect these knowledge sources through two primary methods:
Direct Data Uploads: Directly uploading documents of your choosing to AI Studio to help AI Companion provide more contextually-aware content and references related to user queries.
Third-Party Index Connection: Connecting a third-party indexing solution, like Amazon Q or Glean, that enables AI Companion to retrieve relevant documents and information from various data sources (e.g., Dropbox, Google Drive, Sharepoint, etc.) through the third-party platform.
When using either of these methods, organizations can leverage their internal documents, knowledge bases, product information, and proprietary content to help AI Companion generate responses that better reflect organizational knowledge and help maintain consistency across communications.
For example, when a user asks, "What company holidays do we have off this year?" or "What's our competitive positioning against Vendor X?", AI Companion can either search through uploaded company documents to provide detailed responses or query connected enterprise indexes to retrieve relevant matches and contextual explanations with source material references, depending on the organization's chosen integration approach.
What’s the big difference between Direct Data Uploads and a Third-Party Index?
That’s a great question! In short, Direct Data Uploads lets admins upload content directly to Zoom that’s accessible to AI Companion, while a Third-Party Index Connection empowers AI Companion to query your existing indexing infrastructure to retrieve content in real time. The following two examples provide additional context for understanding this nuance.
Example - Direct Data Uploads
Imagine you’ve given AI Companion a curated library of company documents. If a user asks, “Where can I find the NDA template for external contractors?” or “Where’s the documentation for setting up a new VPN connection?”, AI Companion searches across the full set of documents your company has uploaded to the Zoom web portal. Based on the user’s and the document’s admin-defined access levels, AI Companion can access documents accessible to the user. From there, it uses a large language model (LLM) to generate a summarized, contextual response, along with a link to the source for easy access.
In this scenario, Zoom AI Companion has access to the full content of the documents your company has provided—unlike an indexed system that instead provides AI Companion with relevant portions of text from documents. However, document visibility and responses are still governed by user group permissions, helping ensure that users only receive information they’re authorized to access.
Example - Third-Party Index Connection
Imagine AI Companion reaching out to a dedicated resource—like a company librarian—who can reference any internal documents made available to them. If a user asks, “Who is responsible for vendor onboarding, and what’s the process?” or “Where can I find our competitive positioning against Vendor X?”, AI Companion passes the request to the company’s index (“librarian”). The index searches through the documents it has access to (based on the underlying data the requesting user is permitted to access and noting that not all company records may be included) and, if the user is authorized, returns relevant matches along with a portion of text from each document that’s most relevant to the request. AI Companion then processes that information with a large language model (LLM), generating a summarized response that includes both contextual explanation and links to the original source materials to help the user better understand the answer.
In this scenario, Zoom AI Companion interacts only with the information the company’s index is authorized to access based on the user’s permissions, and receives only relevant portions of the matching documents.
Knowledge Collection Architecture
The following diagrams outline typical dataflows for Knowledge Collections with Custom AI Companion under Zoom’s federated approach.
Direct Data Uploads
Third-Party Index
Third-Party Applications Skills
Third-Party Application Skills: Empower AI Companion to execute tasks across external platforms
Zoom’s Custom AI Companion supports integrations with third-party applications, enabling AI Companion to actively perform work on users' behalf across external systems without requiring platform switching or manual intervention. Through intelligent automation capabilities, AI Companion can execute comprehensive workflows—such as creating, reading, updating, deleting, and searching Jira tickets—while users remain focused on their primary tasks within Zoom Workplace.
This functionality transforms workflow efficiency by enabling AI Companion to act as an intelligent intermediary that helps manage cross-platform operations, centralizes task execution, and eliminates the friction of juggling multiple applications. Rather than simply providing information about external systems, AI Companion becomes an active participant in business processes, autonomously handling routine operations and complex multi-step workflows across integrated platforms.
For example, when users discuss project issues during a Zoom meeting, AI Companion can intelligently analyze the conversation transcript, proactively suggest creating a Jira ticket, and autonomously populate the ticket with contextually relevant details extracted from the discussion—executing the entire workflow seamlessly while helping ensure critical information is captured and actioned without manual data entry or platform navigation.
AI Companion Third-Party Integrations Architecture
The following diagram outlines typical dataflows for using Custom AI Companion with third-party integrations under Zoom’s federated approach.
AI Companion for Third-Party Meetings
AI Companion for Third-Party Meetings: Bring AI Companion to Google Meet and Microsoft Teams meetings
With Zoom’s Custom AI Companion, users can bring AI Companion to third-party meetings hosted on Microsoft Teams or Google Meet. With this feature, users can still receive post-meeting summaries and transcripts, creating a unified, cross-platform AI experience. This eliminates the need for users to switch between tools and enjoy the same AI-driven features regardless of the meeting platform.
AI Companion Third-Party Meeting Assistant Architecture
The following diagram outlines typical dataflows for Custom AI Companion when used with third-party meeting services.
Last updated
Was this helpful?