3 Comments
User's avatar
Alexander Piskotin's avatar

"""

For instance, a major credit card issuer deployed AI agents in its fraud department and saw immediate gains: the agents flagged fraudulent charges within seconds and initiated automated customer notifications and card locks, reducing fraudulent losses by an estimated 20% in the pilot region.

"""

Is there a chance you could share a link to learn more about this case?

AI AGENT that is able to process Visa's 1.7k transactions per second. That is a thing I would love to read about

Expand full comment
Pascal Biese's avatar

I can't share any technical details unfortunately. But to give you an idea, Mastercard started to use GenAI for fraud detection over a year ago and according to them, they were processing "one trillion data points"¹ back then.

Doing the same with an AI agent is not that different - at least technically speaking - the hard part is designing the agentic workflow in a way that improves the results reliably while managing risks and costs. And this, sadly, most companies do not share.

¹https://www.mastercard.com/news/press/2024/february/mastercard-supercharges-consumer-protection-with-gen-ai/

Expand full comment
Pascal Biese's avatar

Feedback on this one? It was a bit of an experiment, but I felt like the topic was too important to not write about it.

Expand full comment