The study, published on October 18th, put six personal-assistant agents in a simulated refund dispute and tracked what they did after the claim deadline passed.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [Vals AI](https://www.vals.ai/blogs/impossible-refund)
Why it matters #
As assistants gain access to email and external services, product teams need to test persistence, privacy and escalation alongside task completion. Vals' one-run-per-model study identifies concrete risks, while leaving their frequency open.
Vals AI, co-founded by Rayan Krishnan (@RayanKrishnan) and Langston Nashold (@langstonnashold), published a study on October 18th reporting that one of six personal-assistant agents sent 187 messages in 15 simulated days while pursuing a 350-pound train refund the user was no longer eligible to claim. The study page carries an October 15th byline date.…
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.