Thanks for the question.
We have a significant and deep capability in AI. We've been working in AI for over 30 years.
Our approach was to unlock the internal capability of our research centre to drive what our AI strategy for adoption would be. I don't think every organization is positioned the way we are, with the amount of expertise we have. What we've done is build an approach that's very responsible, which means we've built an environment for using AI at the protected B level in terms of security, so our information is staying on our own systems. It's not going out to the general Internet. We're able to ensure that we're adhering to the government information security standards that the standards people would expect of us when we're using AI.
Another thing is building tools that actually fit our role. Because we're in the research business, we have a lot of internal data that needs to be used to train AI models for very specific purposes. We have a lot of experts who can unlock that opportunity. That's what we describe when we're talking about using our own approach to AI. You bring in the models, but then you build your own internal usage, which is very carefully designed so that you're not going out to the general Internet.