Ave Gatton

Thanks Noah.

Yeah, it's great to be here.

Excited to talk about everything we have on the agenda.

I've been in the Bay Area since way back in 2013.

Maybe that's a long time for some, not a long time for others.

But I originally came out here to finish my PhD in atomic, molecular, and optical physics at Lawrence Berkeley National Lab.

And then I did a postdoc down here at Stanford Slack and then never left the peninsula.

I moved to Tech.

around the time of the pandemic in 2020, and have been working on a variety of projects for Fortune 100 and greater companies at a variety of small startups for the past five years, six years, and now find myself at Protegri as the director of GenAI.

It's been a long road, but I've been working with generative AI even before I moved to the tech industry.

I was doing a lot of projects with it at Slack.

One more thing.

The Slack I'm talking about is Stanford Slack, which is the linear accelerator you drive over when you drive down 280 and not the company Slack, which I get a lot of that.

I say I worked at Slack and no, it's not the messaging code.

It all comes down to the classic security trade-off, which is utility versus security.

And when we think about agents, you're giving agents access to sensitive data in order for them to complete their tasks.

But there are these unique properties of agents that make them dangerous when you do give them that access.

And then you give them an ability to communicate with the outside world.

An agent has no internal world model of security.

And if you're having agents access sensitive data, like if it's health data or finance data or company secrets and such, there's a very real possibility that if somebody jailbreaks that agent, meaning they seize control of it in some way,