Data portability

Ownership output of what comes out of that system is mine, right? You always want to try to get that.

Data portability, when we're done with this system, I want that data to be mine, especially if I fine-tuned any of the model weights.

They come with me.

You also want to make sure that the system's performing, right? So if you're going to have this system, you're going to maintain it for a few years.

We can certainly create contract language that allows us to do it, but this assumes that the AI system that eventually got implemented and deployed was serving a well-defined business case or use cases that already the contract language was thrown up around.

The challenging part of exiting deployed agent systems, I think that we should look at some three distinct pillars or silos.

One is data portability.

That's the obvious one.

We much prefer a competitive playing field when it comes to shopping agents.

So one thing we would want is data portability.

If you're using, let's say, Google Shopping Agent, Google already knows a lot about me.

Let's say I use Google in all sorts of ways and it also has great, great listings.