Rise of the AI Meeting Assistant

AI meeting assistants are increasingly becoming commonplace in corporate environments, yet much of the world is completely unaware of the technology behind this new breed of productivity tools. These digital assistants are designed to attend meetings, capture conversations and automatically analyze and index the resulting content. AI meeting assistants can help companies find and share insights they may not have otherwise been able to uncover.

To understand the rise of the AI meeting assistant, we first need to understand the underlying technologies that enable it. Speech recognition is a subset of Artificial Intelligence (AI) technology that enables a computer or a device to process and interpret human speech. Historically, speech recognition has only been used in narrow use cases and has offered limited accuracy. However, over the past two decades, this technology has advanced dramatically and evolved from simple keyword dictation to full conversations.

One of the reasons that speech recognition has advanced so quickly is because of the advent of deep learning and related neural network technologies. Deep learning powers the vast majority of commercial speech recognition solutions today. These algorithms do not require engineers to manually program the rules or logic of a system, but rather, they automatically learn from massive amounts of data. As the accuracy and performance of speech recognition algorithms has improved dramatically over the past few years, the number of possible use cases enabled by this technology has also grown rapidly.

AI meeting assistants are still relatively new, but many companies are starting to rapidly deploy these tools based on the benefits seen by early adopters. These digital assistants can perform a variety of functions including automated generation of meeting minutes, note taking, and task management. Automated note taking is one of the most common uses for an AI meeting assistant. Historically, many companies have been dependent on manually generated notes to communicate knowledge and information throughout their organizations. However, note taking is often a tedious and error-prone process. An AI meeting assistant can automatically capture information communicated in meetings and provide a searchable archive of previous conversations. This content can then be mined to find insights, automate knowledge transfer, identify risks, and optimize or automate specific company processes.

These systems typically connect to a company's calendar and actively monitor meeting invites. They then automatically attend all scheduled meetings, capturing the audio and video as well as any chats or files shared by any participants. Once a meeting is completed, these digital assistants go through a process to extract, index, and store the meeting content for future retrieval. This content is then made available for search and discovery.

While this process sounds simple, there are many challenges that need to be addressed to design a robust AI meeting assistant. These systems must be able to consistently understand the language used in meeting conversations. The way we speak can be complex, with jargon, slang, abbreviations, and even emojis. AI meeting assistants must also be able to capture and understand what is being shown on the screen. For example, if a participant shares a PowerPoint presentation, the system needs to be able to understand that slides are being displayed and properly index the content. Finally, these systems must intelligently handle conversations that happen across many different communications platforms, as companies today leverage a wide variety of VoIP providers and video conferencing tools such as Google Meet, Zoom, and Microsoft Teams.

One of the primary benefits of automated note taking is facilitating low-friction knowledge capture and transfer within an organization. AI meeting assistants can automatically capture and store meeting content in a searchable index. This content can then be mined to identify business insights, such as identifying key risks, latent opportunities, or areas of improvement. For example, an AI meeting assistant can help a company identify discussion themes in past meetings by automatically analyzing  which topics were discussed and for how long.

Another use case enabled by this technology is automated knowledge transfer. By providing the ability to navigate and search meeting content, these digital assistants enable knowledge transfer in a low-friction way. For example, a product manager could identify feature requests from a company's sales calls and easily share them with the appropriate engineering team.

Facilitating seamless capture of a company's calls and meetings requires an AI assistant to be capable of integrating with a variety of company calendar, sales dialer, and videoconferencing solutions. These digital assistants must be able to seamlessly integrate with a company's existing infrastructure without requiring extensive effort or causing disruption to a company's existing processes.

An AI assistant must be aware of employees calendar events and the details of those events. For example, an employee may only want external calls and teem meetings to be recorded, excluding internal HR or legal calls. These systems must also be capable of capturing non-scheduled meetings or processing manually created recordings. Support for a wide variety of video conferencing solutions is also critical. Many organizations leverage a variety of tools such as Google Meet, Zoom, or Microsoft Teams, therefore these systems should be able to seamlessly integrate with these platforms.

One of the key capabilities provided by an AI meeting assistant is the ability to automatically generate meeting minutes and other meeting-related artifacts. This is very helpful for companies that want to keep track of actions or decisions that were made during a meeting. It is important to understand that meeting minutes are not transcripts, they are an edited summary of a meeting's content. These systems must be able to capture the essence of what was discussed, not just the words used. The challenge is that capturing and indexing a summary of meeting content is a very complex problem and requires a sophisticated AI assistant.

In a meeting involving multiple participants, a typical transcript can look something like this:

"Person A: So in this release we should focus on improving the reliability of the Android app, what does everyone think? Person B: Sounds good. Person C: Yeah, great! Person C: I'm concerned because our usage numbers indicate iOS is more important right now. Person B: Oh, that's true actually. Person A: OK, let's switch priorities to iOS reliability."

A summary of the same conversation might look like this:

- Team discusses focusing on Android app in the next release.

- iOS usage numbers indicate iOS reliability should be improved.

- Action Item: Focus for next week will be iOS app reliability.

As you can see, the summary includes the same information as the transcript but it is organized more concisely and parses out key details such as action items. It is important when choosing an AI meeting assistant to select a system that can generate high quality meeting minutes and summaries, and not just transcripts.

One of the most powerful capabilities enabled by AI meeting assistants is the ability to search and discover business insights from past meeting content. This is often referred to as "conversation search". When a company is looking for a specific set of information or content, conversation search can be a huge timesaver. For example, imagine that a company wants to find all of the previous discussions related to HR onboarding policies. With an AI assistant capable of conversation search, the company can simply type "onboarding" into a search bar and it will find any meetings that discuss this topic.

Discovery goes beyond search by providing the ability to browse through structured knowledge extracted from recorded calls and meetings in a Wikipedia-style connected graph. The ability to see aggregate views of conversations involving specific topics, people, companies, or products can help companies identify new opportunities, challenges, or changes in the market. Reporting and anomaly detection are also possible by using conversation discovery to compare various metrics with the historical baseline. Is a particular issue coming up more often in sales calls, or is a key metric trending differently in the last few months? These are all questions that are possible to answer when looking at a connected graph of meeting conversations.

Another key capability enabled by an AI meeting assistant is the ability to automatically notify attendees of specific actions or decisions that were made during a meeting. These notifications can include links to the recording of the conversation, as well as any automatically generated meeting summaries and action items. Post-meeting notifications may be sent via email, Slack or Teams messaging, or a variety of other mediums. The goal of these notifications is to facilitate knowledge transfer by ensuring everyone who was involved in a decision is quickly notified.

By providing the ability for employees to digitally comment on specific parts of a meeting recording or create shareable clips of the conversation, the AI meeting assistant can facilitate collaboration within a company. Electronic notes and bookmarks can be created that are automatically associated with specific times in the recording. Recorded and analyzed meetings can form a collective 'ground truth' of what is happening inside an organization, enabling easier customer handoffs, onboarding of new employees, and alignment of team objectives.

Recent trends in the AI meeting assistant space is the move from passive information capture towards intelligent process automation. These AI assistants are starting to move beyond traditional note taking, and are providing operational support. For example, an AI meeting assistant may be able to automate the entire process of filing a trouble ticket, logging a bug, or queueing a task. This type of process automation is extremely valuable to companies that want to free up their employees to focus on more strategic tasks.

In order to automate business processes, an AI meeting assistant must be able to recognize and act on specific key events that occur during a call or meeting. For example, if a company wants to automate the process of filing a trouble ticket, it is important that the AI assistant understand when a customer or employee is requesting help. The challenge is that there are many different ways of expressing the desire for support, for example, "I need help with XYZ," "XYZ is broken," "I'm having trouble with XYZ," or "I don't know how to do XYZ." In order to recognize these different requests for support, the AI assistant must be able to go beyond traditional keyword analysis and use advanced machine learning algorithms to recognize specific patterns of conversation.

Once the AI assistant recognizes a request for support, it must be able to perform intelligent task automation. This can be done in a variety of ways, depending on the needs of the company. For example, the AI assistant might need to connect to a company's ticketing system to file a support ticket, or add a new task to an internal task management tool. The goal of automation is to not only automate tedious work, but also to help the company operate more efficiently and intelligently.

There are many common use cases that are enabled by AI meeting assistants. The following are some of the most common scenarios:

1. Company-Wide Knowledge Transfer

It is often difficult for companies to transfer knowledge from one employee to another. Meeting recordings can be used to facilitate knowledge transfer across the entire company. For example, if an engineering manager wants to make sure that all of their team members are up to date on the customer feedback from the last release, they can go to the AI assistant and search for all of the meetings where the company had conversations with customers. They can then search for opinions, feature requests, or open issues and link to the original meeting recording. When an employee is onboarded to a new team, they can be given access to the entire company knowledge base.

2. Problem Identification and Resolution

By analyzing meeting recordings, AI assistants can detect patterns and trends in company communications. For example, if a company is seeing an increase in calls from customers experiencing problems with their iPhone app, the AI assistant may identify this pattern and alert the company to the issue. This would allow the company to focus resources on improving the iPhone app, rather than randomly troubleshooting.

3. Personalized Customer Support

Customer support is a key area where AI meeting assistants are starting to be deployed. By monitoring all of the historical interactions that a customer has had with a company, it is possible to identify specific issues that have come up in the past. This can help a company provide a better level of personalized customer support. For example, if a customer contacts a company with a problem and the AI assistant recognizes that this is a recurring issue, the system may provide notifications that escalation is likely. This can help a company better prioritize their customer service issues, and provide better support overall.

4. Operations Monitoring and Analytics

AI meeting assistants are also being used for operations monitoring and analytics. By capturing all of the interactions that take place between a company and their customers, partners, or employees, it is possible to generate key insights. For example, an AI assistant may detect when a customer gets frustrated during a conversation, or may identify a salesperson that is not aware of a new pricing structure. These insights can help companies improve their customer service, sales, and other operational metrics.

Over the next few years, AI meeting assistants will continue to mature and provide new capabilities beyond traditional note taking and knowledge transfer. These systems are starting to move from entirely passive roles in calls and meetings to more active roles. This includes meeting facilitation, the ability to actively guide participants and suggest next steps, as well as process automation of tedious manual tasks. As more companies begin to adopt automated AI meeting assistants, the ability to use these systems to share knowledge, collaborate, and plan business operations will increase exponentially.

While many people may not be familiar with the term 'AI meeting assistant', these systems are now prevalent in many major enterprise companies. As these AI assistants mature and enable companies to capture, share, and automate business processes, the benefits of bringing AI into calls and meetings will be even more obvious. Companies that want to embrace AI should start thinking about how to enable these systems, and what new capabilities can be added.