Hacker News Reader

> Abanca is using agent orchestration to handle sensitive customer information at a huge scale (2 million customers in their app).

Maybe my perspective is skewed on what "huge scale" means, but 2 million users? That's like a few hundred megabytes of data? Or a couple GBs if there's a lot of per-user data?

View on HN

Maybe, but using state-of-the-art large language models to solve customer support queries with agentic can quickly use a lot of tokens. What I understood from the talk is that they used agents with limited responsibility and (assumption from me) smaller models, to the make sure the answers were quick, reliable and not too costly.

There are several payments processing companies that are already largely using AI for customer support queries. They still have an escape hatch to a human but at least one of those companies (on the smaller side) is reporting a ~99% success rate, they are down to a handful of human customer service employees now for cases where the customer can't find/produce the transaction ID.

Using Mistral or more SOTA models?

European consumer focused businesses do not scale easily the same way US ones do, which is a major contributor to their problems developing tech businesses generally.

OTOH such things can be quite defensible, they just rarely become anything like as profitable.

Reply on news.ycombinator.com