One AI agent can't catch everything in high-stakes decisions. A better approach chains multiple agents with different jobs: a strict fact-checker, a lenient reviewer, and a third trained on your company's stance to make the final call.
That's again, where I think the world of you keep hearing all these things about agent, agentic web and all of that. It's like an agent basically has some outcome, but that outcome can be passed to another agent, which then has a different objective and a different outcome along the way.
You know, you're. You're working on these with our team, developing agents day in and day out, but I know you're also using some different ones and looking at different things different ways and researching more than I am. But would. It just came to my mind. It almost seems like you could put, you know, in the, in the fact check and maybe in multiple spaces of that path you just described, where you had agents that were highly critical and then others that were maybe one that's highly critical and one that's maybe a little more lenient. Have them both come up with an answer and then have an agent that's trained on your true view and your company's position. And maybe those are separate agents, right, that. Then take another look at it before you get an output.
What does AI actually mean for a wholesale distributor #WholesaleDistribution
If your systems don’t talk to each other, your AI won’t either #AITransformation
