Okay, that title’s a bit misleading. The product doesn’t exist. But I wish it did. And if you’re subject to many meetings, you do too. So let’s figure out how to build it.
Here’s my thinking. Often in meetings, much conversation contains no relevant information. This is because attendees often:
- Repeat themselves, or repeat other people
- Go off topic and have tangential conversations
- Break, such as for lunch, coffee, or the bathroom
- Use stop words and filler content, and otherwise don’t get to the point
- Posture and signal status rather than talking substance
My guess is that, using the Pareto principle, 80% of a meeting’s value comes from 20% of the content. If true, then a compression algorithm addressing the above could deliver 80% of a meeting’s value in 20% of the time.
I have some ideas for how to do this. You could train a machine learning model to identify 3, 4, and 5. You could use statistical techniques to deduplicate repetition for item 1. And you could use natural language processing for number 2. (You could use it to understand the core topic of a meeting, and when people strayed outside of it.)
Granted, you couldn’t compress a meeting while participating in it. This would only work for after-the-fact analysis. But you could use a system to flag items 1-5 during the meeting itself, and keep people on track.
Let me know if you’re aware of such a product. Or think about how to build it in your next meeting that could use some compression.