And it's also important for optionality, right. If you if you want to be able to pivot your strategy and have that flexibility going forward, you like to to know that you have the option to bring different data sets into a platform while maintaining that fact scale at the core, I think the I use case is, evolving, through a number of stages.
Right. So I still see most of the emphasis at the moment on the productivity use cases. And certainly that's where we're deploying AI the most. There's huge productivity gains in terms of, you know, data validation in terms of automating tasks that are previously manual, ingesting data from non structured sources, things like that. But I think that's evolving over time.
I think the next frontier in the medium term and some of this is coming to life now is workflow embedding of AI. So things like having best execution order routing, using AI for some of those sort of use cases, and then the final frontier, the eventual end state is this, deployment of more agent AI to potentially replace entire workflows so that your user experience and the workflow tool is basically an AI agent and you're just, you know, prompting it, making suggestions in terms of where those investments may want to be deployed, and then allowing the AI to actually perform the workflow for you.
And that's a bit further off, but we're enabling that as well by trying to build that data foundation, share data to the client, and allow them to begin experimenting with deploying agents within a safe sandbox environment. So we think that, you know, today, productivity is still the number one focus, workflow is the emerging piece. And then the end state is this agentic approach.