How to Train an AI to Know What Needs to Stay in a Meeting

Artem Koren, co-founder and Chief Product Officer at Sembly AI, started the company in January 2019 to bring the power of AI to online meetings. 

Artem and his team developed an app that listens in on virtual meetings and does all the note-taking for you including recommending action items and suggesting the most important topics. These are hard AI problems to solve and Sembly’s success is an indication they’re off to a great start.

Before Sembly, Artem was an executive and co-founder at companies including Neusana and Visual Trading Systems and he spent time as a manager in big company land at Ernst & Young.

Listen and learn...

1. Why Artem and his co-founder decided to fix the problem of broken meetings 
2. Why the evolution of online meetings… is like the evolution of airplanes 
3. Why we’ll soon send AI agents to attend meetings on our behalf 
4. When meetings are required… and how to make them more efficient 
5. How neural nets are solving traditional voice transcription problems related to accents and background noise 
6. How to solve the problem of automatically determining who said what in a conversation 
7. How Sembly uses generative AI to summarize meetings 
8. What are the risks of having AI decide what tasks to assign to meeting participants 
9. How to prevent sensitive information from being passed to large language models as training data 

References in this episode... 

• Safety and ethics are being compromised in the rush to get new generative AI products to market
• Krish Ramineni from Fireflies on AI and the Future of Work
• Rich White from Fathom on AI and the Future of Work
• Sembly AI