Read AI, a Seattle startup that applies artificial intelligence to online meetings, is expanding beyond its automated recaps of single meetings — using AI to summarize and glean insights from potentially thousands of meetings at a time.

This broader approach, which ReadAI calls “Large Meeting Models,” or LMMs, is the basis for several new features that the company is releasing.

Read AI, the Seattle startup whose software measures engagement and sentiment of participants on video meetings, is taking a page out of ESPN’s “SportsCenter” playbook with its latest product.

Much the way a sports highlight reel distills a complete event into a more manageable and entertaining chunk of video, Read has launched “Read Highlights” to convert video of a work meeting to find the best parts.

By leveraging OpenAI’s GPT technology, Read Highlights can convert a 60-minute video recording into a 2-minute highlight reel. Large language models integrated with video identify the key moments from a meeting, analyzing not just the spoken words, but reaction from participants.

Read Highlights converts a 60 min recording of a Zoom into a 2-minute highlight reel.

By integrating large language models (GPT) with video (in-house), Read identifies the most important moments in a meeting based on not just the words, but the reaction from the participants.

Read uses visual AI to measure reactions from meeting participants to highlight the most and least important moments in a meeting to generate Read Highlights.

Read combines OpenAI's GPT's language models with their multimodal audio and video models that measure meeting participant engagement and sentiment to create a novel approach to meeting summarization.

The output is not just a transcript plugged into GPT, but text with annotation around how did non-speakers react to what the speaker said.

he modern meeting is broken, yet we see a 153% increase in meetings over the last two years. We’re piling bad meetings on top of bad meetings, with the hope that things will magically change using medieval solutions like “No Meeting Mondays”.

In the last week, several studies brought light into the disconnect between employees and managers as it relates to productivity and its impact on the number of meetings that we attend:

* 85% of managers worry they can’t tell if employees are getting enough work done, while 87% of employees say they are productive

* The CEO of Microsoft stating that, productivity paranoia results in actions like spying on employees

* The concept of productivity theater, where employees prioritize actions that look like they are working versus actual work

* 1 in 3 employees accepting and sitting in on meetings where they know they should be a participant

This type of analytics will make online meetings better in the long term, in the same way fitbit measured steps to quantify are you walking enough.

A recent study (Q122) from Read highlighted (https://www.read.ai/benchmarks) that 28% of meetings have unbalanced participation and that 11% of participants in a meeting are in "ghost mode", no camera, no audio. The more measurement, the more opportunity for individuals and teams to improve.

If you utilize Google Analytics, Flurry, Mixpanel, etc... this is a straight forward addition as a free service to measure location down to real-world places.

Interesting approach with mobile analytics, focused on location, versus the standard in-app activity that a Flurry, Mixpanel, or Google Analytics would measure.