Jyunmi Hatcher

the authors had reported.

The trial enrolled 5,325 patients age 65 plus and delivered results to clinicians inside the EHR inbox, avoiding extra visit time or license fees.

Because the AI runs on data already in the chart, it can scale in lower resource settings, though external validation bias checks are still needed.

The study offers rare RCT evidence that AI plus a brief patient survey can raise real-world detection without burdening clinicians.

Exactly.

And it does sound like this is an application of edge AI or smaller models because it lives within the record system itself.

Now, need to get more details to see how those systems are within any health organization or within the clinic.

But if we can scale down the AI to be attached to that one locale, that does help with things like HIPAA and whatnot.

I I'm, I'm sure they will want test subjects who have very focused minds.

So that puts me out there.

Yeah, yeah.

I want to talk about Time Magazine.

So Time debuts their AI agent, and it's trained only on its 102-year archive.

Nice.

I didn't realize Time was around that long, but apparently.

So Time launched an AI agent that answers questions and generates text summaries and audio briefings from roughly 750,000 items in its own archive.

The publisher said the system built with scale AI supports 13 languages at launch and is limited to time content for answers rather than the open web.

Time frame the move as engagement first, not immediate monetizations.

And leaders said that the agent will later gain logged in memory and may expand beyond politics and entertainment into more of the site.

And it is available to use right now.