Meetings – a necessary evil in the world of work. No matter how you slice it, they are time-consuming and there seems to be no way around it. According to a recent poll conducted by Deputy.com in 2022, employees in the U.S. spend an average of eight hours per week in meetings. Of course, this number can vary depending on the industry and location.
The unproductive nature of meetings has led to the increasing popularity of AI-powered summarization tools. In a recent survey by The Conference Board, a non-profit think tank, nearly half of the respondents stated that they rely on AI to summarize the content of emails, conference calls, and other work-related communications.
While some videoconferencing platforms now offer built-in summarization features, there is still room for third-party solutions. David Shim, co-founder of Read AI, believes this to be true. Read AI is a platform that specializes in summarizing video calls across various platforms such as Zoom, Microsoft Teams, and Google Meet.
Shim, who was previously the CEO of Foursquare, co-founded Read AI in 2021 alongside Rob Williams and Elliott Waldron. The trio had previously worked together at Foursquare, Snapchat, and Shim’s previous startup, Placed (which was acquired by Foursquare in 2019).
“Read AI’s direct rival is traditional project management, where notes are manually taken,” Shim explains to TechCrunch. “But with our platform, we go beyond being a mere co-pilot. We act as an autopilot, delivering relevant and valuable information that increases efficiency and effectiveness in the workplace.”
Initially, Read AI focused solely on summarizing video meetings by offering dashboards for measuring the success and effectiveness of these meetings. Along with this, they also provided two-minute summaries for hour-long meetings. However, following a recent funding round of $21 million led by Goodwater Capital and Madrona Venture Group, the company is now expanding into message and email summarization.
This new capability is currently being tested in a “soft launch” phase and is designed to connect with popular messaging and videoconferencing services, including Gmail, Outlook, Slack, and more. This allows the platform to learn about relevant topics and deliver daily updates with summaries, AI-generated “takeaways,” overviews of key content, and conversation topic updates in chronological order. The service is priced between $15 and $30 per month.
“What sets Read apart is its ability to connect and coordinate various forms of content through AI agents working in the background,” Shim emphasizes. “This unified approach to communication empowers individuals and their teams with personalized, actionable briefings tailored to their specific needs and priorities.”
Of course, there are some who may question the accuracy and consistency of AI-driven tools in summarizing content.
Models like ChatGPT and Microsoft’s Copilot have been known to make mistakes when summarizing, often due to their tendency to “hallucinate” or insert incorrect information. The Wall Street Journal even reported a case in which Copilot invented attendees and fabricated topics that were never discussed in a meeting.
But Shim argues that Read AI is more reliable and robust compared to other solutions like Supernormal and Otter. “At Read, we have a proprietary methodology in place that coordinates raw content with outputs from language models, allowing us to detect and correct any deviations automatically,” he says. “Additionally, we also use content from meetings to provide context for emails and messages, resulting in more accurate summaries and better results.”
It’s important to take Shim’s statements with a grain of salt, as he has not provided any benchmark results to support his claims. Additionally, there have been concerns about Read’s sentiment analysis tool, which uses facial and vocal cues to analyze participants’ emotions in a meeting. This has raised issues about privacy, bias, and data security.
Multiple studies have shown that AI algorithms can be biased towards certain races and genders. For example, emotional analysis models have a tendency to assign more negative emotions to Black individuals and perceive their language as aggressive or toxic. And in a study conducted by MIT, it was found that algorithms could become biased towards certain facial expressions, resulting in decreased accuracy.
Despite these concerns, Read’s growing customer base and recent funding round of $32 million indicate that the platform is here to stay. Shim plans to use the investment to double the staff to over 40 employees by the end of the year and continue expanding the capabilities of Read AI.
“Despite the slowdown in recent years, Read has seen a steep increase in users, meetings, and revenue,” Shim concludes. “This growth can be directly attributed to the significant time savings that users experience when using our platform.”
In the end, only time will tell if AI-driven tools like Read AI can truly deliver on their promises. But for now, they continue to make strides in the world of business, reshaping the way we approach meetings and communication in the workplace.
