Yeah.

And so with Cadence, um, uh, the physician orders Cadence.

Cadence gets the patient a cellular connected blood pressure cuff, a scale, um, devices that give us their vitals remotely at home.

The patient starts taking their ru- vitals.

We have their full medical records, their labs, vitals, allergies, symptoms, everything.

And we're using AI to figure out, is this patient on the right drugs? If they're not on the right drugs, let's prescribe new medications, adjust current dosages, remove old medications.

And we do that with all in an automated fashion with humans in the loop making the final decision on these med changes.

Okay

I

mean, I would say 2022, when the chat GPT came out and the first LLMs became available, it became pretty clear to us that we were going to be able to accelerate our vision very quickly.

And so we focused and invested pretty heavily on starting to experiment and build on top of these large language models ourselves.

And the places where We're now almost four years into that journey, and the places where this has really proven to be valuable, I would say three areas.

One is understanding the patient.

So these chronic disease patients that we treat are on average 74, 75 years old.

They have very long, complicated medical histories, multi-comorbid, heart failure, diabetes patient, for example.

less savvy, I guess? Or have you seen something different at your populations? Or do you do things to make it easier for them to like participate in this type of care model that's more autonomous?