And we deliver, based on our own technology, an LLM as a judge, an LLM that's evaluating the answers of the other LLMs, that it's basically guaranteeing that every output that we generate, it's quality-based and there is no hallucination.

Yeah.

And one thing just in terms of practical terms, in terms of the retrieval orchestration.

So without retrieval orchestration, if I'm just asking a question of an LLM, say I'm a salesperson, I say, what's happening with my account? Without RAG, you're going to get information that's based on or if you use Claude or ChatGPT today.

you're going to get information that's based on the web, based upon their training data.

And so what we're trying to do with agentage orchestration, so if a salesperson asks, I'm heading into a meeting, I want to know what is going on with this opportunity X at this company, what the retrieval engine can do that now with that information, is it can say, hey, I want to go into my enterprise knowledge base and maybe pull up some competitive differentiation.

So it splits the question into sub queries and tries to understand how best to fulfill that answer.

What stakeholders are collaborating in here, and how do they do that?