All right, so Rob, when you first started working with Mercy Corps, well, we talked a little bit about how you got introduced, but what problems were presented to you and what was the mission that you were involved in and how did you come to figure out that AI might be able to help?

Yeah, I mean, Josh has actually highlighted quite a few of the problems which we were trying to address, which is the disparate amount of data or disparate data sources that existed because the crisis analysts that were producing reports on certain situations or crises, they were spending a lot of time going through all these different humanitarian APIs and portals and social media and looking at local news sources and like it's a never-ending and ever-increasing and ever-growing list of data sources and it just took so much time that you know from like josh was saying by the time you finished all that analysis or just the information gathering the situation has changed already So it's how do we get to the decision side quicker and make that more actionable? I mean, AI seemed like a great fit because of the complex and varied nature of the type of work and queries that were being asked by the analysts.

We let the agents decide, do I need this humanitarian API for this specific query? We have different...

domain specific agents for like conflict or displacement or or local media sources and we let that orchestration happen within the solution to say well you know we we need to check for local sources or this is a crisis domain question so we we make sure we're probing the right apis for the right situation the right scenario and it's it's it's it's different because you can sort of query it's multiple turns in the conversation so each step in the conversation might bring in a different api based on the updated conversation thread so it's like an ever-evolving sort of scenario that the analyst gets to plan in the solution

But Rob, back to you, just in terms of responsible AI and governance, what were some of the challenges and lessons as we think about responsible AI?

I think everything Josh said is the kind of things that we at Clare d'Era as a company always push forward as messaging with our platform, private AI and data sovereignty and all these kinds of not restrictions, but responsible boundaries that we put on these data architectures.

But I think Just adding or expanding on that, I guess, the explainability of AI.

It was from the very get-go when we were designing the solution, it's every source needs to be cited.