Navrina Singh

Are they going to perform really well on the use case that I care about?

Are they less costly for my business, etc.?

There's a lot of things that go into making sure that

that a model actually works for an application.

And this is where Credo AI's governance platform and things like Model Trust Core come in to make that informed decision for a business.

Great question, Corey.

So I've been very involved with the policy ecosystem in the past, especially five years, because a core thesis we had was and continue to have is you will need a very, you know, you will need a core trust layer in this AI stack to operationalize.

not only all the standards, regulations, guardrails that are coming out from policy ecosystem, but they also have to work for the data and AI infrastructure.

So how do you actually create that trust layer that basically does that codification?

And so having worked at the intersection of technology and policy now for five years, the reason I believe this action plan is a moonshot moment is because it is...

you know, really putting a stake in the ground in terms of how United States is going to win in this race for AI.

As an example, there's a pretty big focus on open source innovation.

That is something we have actively advocated for the past few years, because the best way to build compelling, competitive technology and is going to help us continue leading in United States is by bringing in collective wisdom of researchers from all over

the ecosystem.

And I think that commitment to open source is a massive one.

The second thing which I'm really excited about is the entire theme in the AI action plan is grounded not just in technology capability, but in trust and governance.

And I think what is really exciting to see is

very clearly codifying the evaluation ecosystem.

What does good look like, whether it is coming from standards, whether it is coming for sector-specific metrics and evaluations, that has been pronounced and clearly stated.

Again, something Credo AI has done