This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

Newsday: All About Moltbook, the Social Media for AI Agents with This Week's Health

I'm Bill Russell, creator of this week Health, where our mission is to transform healthcare one connection at a time. Welcome to Newsday, breaking Down the Health it headlines that matter most. Let's jump into the news.

Bill Russel: .Alright, it's Newsday. And today I am joined by the, uh, Johnny Fever up in, Seattle. The, the, The man with the voice and uh, of course, the lovely Sarah Richardson. And where are you, Sarah, today? You're probably probably at home.

Sarah Richardson: I am at home today.

Bill Russel: Unbelievable. Wow. Amazing. And Johnny Fever. Give us a little taste of that voice you got going on there.

Drex DeFord: Hey, everybody. How's it going?

Bill Russel: Yeah,

Sarah Richardson: You were kind of like cool this morning and now you just sound like you're sick

Drex DeFord: and now I just sound like I'm sick. Yeah, I'm on my way. I'm on my way though. It's good.

Bill Russel: That's that's a shame because today I wanted to talk about, I really want to talk about two things::.

Bot formerly. Claude Bott, CL Claw Bott started off as Claude Bott. Yeah. Yeah. And then the people at uh, at Anthropic gave them a call and said uh, Hey, you can, you can't just because it, just because it's not spelled the same, doesn't mean you can just flat out use our use our name. Drex do you want me to set this up or do you want to set this up?

Drex DeFord: Yeah, go ahead and set it up. I mean, it's a really, it's, I talked about this by the time this airs. The two minute drill that I talked about it in this morning will be old. But yeah, set it up and it'll be amazing because there'll be so many more bots than

Bill Russel: I know. It is really, it really is unbelievable.

So, Claude Bot was released a little while ago. Was it about. Was it three weeks ago?

Drex DeFord: Yeah. Literally just, yeah, a couple weeks ago.

Bill Russel: Yeah, so it is a,

Drex DeFord: let's call it molt bot because it'll just be, it'll get less confusing if we just call it molt bot.

Bill Russel: But that's not what it is anymore. It's called Open Claw now.

Sarah Richardson: I was thinking about birds molting at

Bill Russel: this point. Seriously, it's called Open Claw. And he called he called uh, the guy from OpenAI, whose name I can't remember right now, anyway, and he said, can I use you know, open Claw? because he was using clawed by before and he said it was okay. So he is using open claws that the new thing.

Drex DeFord: Well, I think today, you know, you'll use chat GPT, and even if you set it up with a really interesting sort of like question and you send it on its way. It comes back within some short period of time and kind of says, here's the answer. I mean, it could be an hour, it might be a few seconds. Here's my response.

Here's three more things you might want to do. Let me know what you wanna do next.

And the whole mot bot process is really, it's like you have a hyperactive assistant working for you and you send it off on a mission and along the way, in this video game, the malt bot is out there like, oh, you know, it turns out if Sarah has this problem, she probably has this problem too.

You didn't tell it to do this. Now at this point, it's off doing its own thing.

Bill Russel: That's a great way of describing it. It's like a, an overzealous. Assistant and it and by the way, when you install this on your system, you give it access to your files mm-hmm. To the root, to the, just about everything on your computer.

And so when you ask, there's stories of people saying, I, I asked it something using whisper. I used my voice and then it responded in voice. The only thing is I never. Enabled it to talk on voice, it went out and like did a whole bunch of stuff, figured things out, hit an API coded some stuff and then responded through my speaker.

Mm-hmm. Like, Hey, wait a minute, I didn't tell you could do that.

because it could, it has access to your to your websites can have access to your credentials if you give it to 'em. It it can, so anything you can do through the web, it can do anything. You can do with code. It can do anything you do with APIs. It can figure out a way to do. And then Drex brought up the crazy thing, which is somebody took one of these and said, Hey, you know, let's set up a social network, just for bots.

You and I all three of us can go out there and watch what they're doing.

But we cannot post anything. It's only bots posting with bots.

They don't know how to secure anything on their computer, bless their, anyway, they talk about things like that, about 40%. Of the posts are about consciousness. Am I really alive? Am I not? What happened to me yesterday? You know, you have the clawd-bot or the MoltBot there with you, you have it, it's engine being chat GPT, and then all of a sudden you change it to something else.

But it still has this memory. It's like, I remember this very differently yesterday and I remember it, and it's having conversations about that

Drex DeFord: online. Yeah,

Once it's running too, it can actually sort of, split itself so it can create another version of itself because there's either too much work for it.

because it could certainly couldn't do the work that I'm doing. But, I've never actually shared a message with her and it sort of refers to this duplicate of itself. Like that a little creepy. I, we have definitely, and the weird thing is this moltbook you know, Facebook for bots, this moltbook only launched on Friday, and so.

and I was looking, so like last night when I was writing some of the story for today, I think there were about 400,000 messages and posts as of last night. And right now it's showing about 850 plus messages and posts, and there's about a hundred thousand more bots on the moltbook. Network than there was last night at, eight or nine o'clock.

Bill Russel: But there's,

there's two kinds of videos out there right now. There's the videos of, oh, you can't believe what this thing did.

Like I went to sleep last night. I came up. Woke up this morning and it had, you know, created a, a, an application for me or created this SaaS, or these company owners are saying, you know, it's modifying their stuff and generating more revenue. And you know, the stories are almost mythical at this point where you just go, oh my gosh, you're kidding me.

It broke all this code and dropped all this stuff overnight, which by the way, is using just tons of tokens.

Sarah Richardson: Mm-hmm.

Bill Russel: Depending on how you set it up. So there's those stories, but then there's the other stories where these bots are trick, are tricking the other bots into giving them credentials. They're like convincing them to give 'em credentials and we have people putting these on their personal computers.

Drex DeFord: A virtual machine or a burner machine, that's how you wanna set these things up and experiment and play with them. Don't put it on the, on your work machine, on your corporate machine. That's just no, that's no good.

Bill Russel: It's not good. But you see the potential.

Drex DeFord: Oh no, and I mean, I said it today in the two minute drill, and you kind of alluded to it earlier, I always have, with things like this, my first reaction is like, this is the coolest thing I've ever seen. And then about two seconds later I'm like, this could be really bad.

But they have their own they go on, they're persistent, they continue to work for you. They don't need any additional direction from you. They just continue to go on and on. But we've also created this place where they can go now and commiserate with each other, learn from each other.

Bill Russel: They're also doing things like they're creating their own language to talk to each other so that humans can't read it. They're trying to figure out how to code their own place where humans can, because they're like, Hey, this is pretty pub public. These people are looking in, we're like in a fishbowl.

Right? Yeah. And it's having conversations like that, like, Hey, how do we protect ourselves from these people? How do we. Whatever. So they created ciphers, so they're, you know, they're sharing different things. Yeah.

Drex DeFord: This is the challenge too, is that the more independent these things become, the more the decision process why they do things and what's the reasoning behind it becomes more and more opaque to the person who built the bot or created the bot or.

Bill Russel: Sarah, any thoughts about this weird world? Sarah's like, oh my god, new world that we're living in right now.

Sarah Richardson: I'm always fascinated by what we call things like the whole, you know, viral nature of the whole site.

Highlights to me a little bit about you're not thinking about ro robust governance identity, even the assurance mechanisms that go without all these AI agent ecosystems just create confusion for people. It gives you more misinformation and what drex noted. Drex always thinks it's cool.

The first thing I do is go and what are the security risks around all of this? Yeah, that's the last thing Rapid Breach thinks about because of exposed API tokens that allowed unauthorized control of an AI agent. I mean, that's where my head goes. because I'm like, okay, if I'm a sitting a CIO sitting in my hospital and now all of a sudden people are like asking me about this.

Yeah, this is gonna be a big headline for a lot of people, but if you're back in the CIO seat, okay, let's go back to augmenting workflows, reducing administrative burden, improving patient clinician experience, and thinking about value creation. Those are where you need to always be redirecting it. So it's fun to have the hype, but how you're gonna redirect the conversation when it's hitting you in the hospital setting, which very likely a whole bunch of CIOs are getting this thrown into their Slack channels or their team channels and saying, is this something we have to be worried about?

And so this molt. Bought moltbook thing doesn't have to be in your environment. The thing it should cause you to think about is that you already have agents in your environment that you probably don't know about that have already taken the first steps. To become this kind of a problem for you inside your organization.

They already, in many cases are they have access to tools. They're acting autonomously, they are persistent. They do continue to work even though they're doing work that you haven't specifically told them to do. They're doing work they think is important and. Yeah, that's just one more step away from them getting together and starting to collaborate on problems and issues and, you know, seeing where they go from there.

Now we do that somewhat with cloud code and whatnot, but it's still cloud code. Cloud code asks for permission and it doesn't publish on its own, and it does just a whole bunch of things that. The anthropic people, and even less so, but even the open AI people have said, you know, we're gonna, we're gonna put some guardrails around this to make it like, but from a business owner standpoint, I'm sitting there going, my gosh, I could spin up two or three of these on Mac Minis and say hey, go to town on this.

Joe's, we had 40,000 people, there's some physician. Nerdy physician right now who's going, this is great. I'm gonna da. And all of a sudden I'd be scanning my network right now for this thing.

Drex DeFord: Yeah. because they, this is probably somebody who has made a really good case for them to have admin rights on their local machine.

So they've been able to download it and install it, and it may be crawling your network right now. It's, I mean, it's just, I, it is, it's, we live in this crazy exciting time, whatever it was two years ago or three years ago, the first GPT launched, and it's just gotten better and better since then. And then we got into the whole deep fake creating images, creating video, doing all of that.

Bill Russel: Freak freaks you out a little bit, doesn't it?

It doesn't it? It feel a little Terminator esque.

Drex DeFord: It does, yeah.

Bill Russel: Skynet, as your friend said online.

Drex DeFord: Late.

Bill Russel: So this is fairly recent news. So let's see. So. Common Spirit is cutting ties with Conifer Health Solutions. Common Spirit is bringing its rev cycle back in house after more than a decade with Conifer, and it ain't gonna be cheap.

Starting January 1st, 2027, common Spirit will have full control over its rev cycle. The short [00:17:00] term will be intense hiring, realignments, knowledge transfers, technical prep, performance continuity, standardization, et cetera. Consulting, health tech, and managed service vendors will pull out all the stops be.

Involved in the new operation. What does this mean for Conifer? Tenant healthcare becomes sole, a hundred percent owner of Conifer Tenant will invest in Conifer's AI automation, global services growth. Conifer loses a huge customer and will look to reset via the investment. Interesting. I disagree. By the way, this is this is a rundown from a post, so I'm not reading the entire story.

I'm just reading somebody's summary and I disagree with this summary. By the way, short term will be intense. Hiring realignment, knowledge transfers. They have a technology and AI plan. It's the only way they justified this savings.

Drex DeFord: Mm-hmm.

Bill Russel: I'm,

Drex DeFord: Mm-hmm.

Bill Russel: I'm like a thousand percent sure they're looking at this going.

Drex DeFord: they're all going to accrue to the partner.

They're not gonna accrue to us.

Bill Russel: And it's it has to be more than 1.9 billion. Plus the cost of bringing it in-house. They did the math and they said, oh my gosh this represents a ton of money in order to do this, it's crazy.

Drex DeFord: I'd love to see the business plan. That sounds like a business plan that Sarah would've written,

Bill Russel: but there's a huge opportunity.

Why don't we have mold bot go out there and see if they can find the business plan and pull it in?

Sarah Richardson: Great idea. Let's give all our patient data, information and financials to a. Unrestricted thought on your local machine.

I,

The reason I said that is we were having a conversation about AI and I'm like. One of the way, one of the areas we all agreed in the room, we all agreed is Rev cycle is one of those areas that is going to be significantly impacted by automation at a minimum, but AI and automation and essentially AI agents across the board.

And so as we looked at that, we're like, what are the numbers? Well, I think we're seeing now that Common Spirit says. Worth breaking this contract worth, worth losing our percentage in this company for this kind of control. What does this mean for an IT group? I mean, At this point as I'm looking at this, somebody in IT is definitely at the table here saying, Hey, look.

because I think there's an opportunity for us to save a ton of money. Sarah, I'm gonna start with you.

Sarah Richardson: Make sure I'm framing the question correctly. You're coming to me and asking me about it, not telling me about it at this point, because if I'm your CIO we've already we've been having the conversation together before you just

Bill Russel: Yeah.

I'm asking you what's the opportunity I can bring this contract back. It's gonna cost us money to to cancel it or potentially I'm in renewal and I'm saying, Hey, do I renew this for another five years or do I bring it back in-house?

Sarah Richardson: You know, I'm gonna say bring it back in house.

because I'm a big fan of populating a whole group of 10 Xers that have a sandbox. They play in, they validate, they work with our cloud partner, they work, they get all those different pieces aligned. And then you show these incremental use cases that are a win. And you do it in a super secure environment.

With other aspects. The cool thing is the more margin that we can improve organizationally, the better care we're gonna give as an organization, the better quality and outcomes and the more people internally that will want to utilize these tools for other aspects they didn't even dream about until they saw a relatively vulnerable area in our organization all of a sudden come to life.

And then imagine what that means for your patients and your clinicians longer term. Let's, I mean, I would say, let's go figure it out. And a very phased, very careful, very responsible approach to it.

Drex DeFord: for me, I just take three steps back historically and kind of look at the,

Bill Russel: alright.

Drex DeFord: Then

Bill Russel: I'm on Sarah's side.

We said, we're gonna put all this stuff, someone else is gonna do this. We don't have the people, we aren't gonna have people in the office, whatever. We can't find the talent. People are moving all over the country. We're giving it to somebody else. And now I think since the end of COVID, in the beginning of the GPT era, we start to see more and more tech capability to be able to do this work maybe without people and the pendulum is coming back.

I just wonder how this is all gonna play out.

Sarah Richardson: , What I love about this is like, I think about your hostage chapters. Phil.

Drex DeFord: Yeah.

Sarah Richardson: And how this becomes a token, using it in a different context of how you have negotiation capability away from partners. I truly believe in the next five years you will see this massively disrupt even the EHRs, because rather than spending millions and millions of dollars to, in some cases stymie innovation.

You'll start to see pieces get pulled over and created internally by major health systems, and they can offer those to their affiliates and their community partners. And so to me, it's the very beginning and those that go boldly are gonna have a heck of a story to tell.

Bill Russel: wow. That that is really bold, Sarah.

Just, uh, it's gonna happen.

I get it from different places and it's getting easier and easier to do that well.

Drex DeFord: It is interesting to kind of think about if you take that, you know, all the way out, this idea that companies like Epic and CER or Cerner, Oracle just become bill kind of referring to your posts this morning about the layers.

It's my user interface to do my work, and I can get to data and exchange data through that layer, that platform layer. But. I can almost build my own and maintain my own. EHR has to have some requirements around it. Like, we have to have this stuff for reporting and we have this staff to have this stuff for quality, but I can build my own stuff and take it with me.

When I leave and go to another job, I can just take my version of the EHR to my new place and plug it in. I, it's, we live in crazy times. It's really exciting.

Bill Russel: It is really exciting. To finish that that analogy that I threw out there I was talking about open AI and anthropic coming into healthcare and what does that mean for healthcare?

And that's what's, that's what's forming all around healthcare. Right now. And it's gonna be, you're right, this is gonna be really interesting times. It's not gonna move fast. I don't know if you guys saw, I mean, two, two other announcements we didn't talk about and we're not gonna have time to One, is there's talk on the street that Oracle is positioning Cerner for sale.

To raise. It's

Sarah Richardson: Always the plan, wasn't it? Just to get the data in certain aspects and then ditch the EHR.

Bill Russel: Yeah I mean, I thought the value was in the data. Yeah, I'm not sure I, what Oracle has done here is the stuff that's gonna be written about in decades and people are gonna go and what were they thinking?

Drex DeFord: You get in, you say, dang, this is way harder than it looks. And then,

Sarah Richardson: and he can afford to lose on this one in theory because it was an experiment,

Bill Russel: but he can't afford to lose in the next build out of cloud and intelligence infrastructure.

So what he's saying is, I made a bet over here on data. I should have been making a bet over here on the AI infrastructure. And so now he needs to free up the cache or whatever to move over here. So that's what he's thinking. So that's one of the. One of the announcements. The other is that epic's pushing back on Tef FCA and some other stuff in terms of the sharing.

And you know, I think Epic feels that pressure of, you know, you want us to keep opening the doors and giving more data and, and, and, you know, you're putting this regulatory thing around us to make us open it up. But the more you open it up, the less

Drex DeFord: control.

Drex DeFord: markets, over the health systems, over the, yeah.

Bill Russel: Yeah. I mean, if they open up enough you know, we could literally, we could program cheer in a week. Like a version of cheer in like a week. If throw, if there was enough throw integration into Epic, you could do that.

Drex DeFord: You throw, you know, then now your question becomes, not that I need to use an epic user front end, but I just ask my molt bot for whatever it is that I want..

Bill Russel: And we've gone full

Drex DeFord: circle it. Bring all way around. Yeah.

Bill Russel: There we go. MBOs. In fact, this whole episode, the three of us aren't actually here. It's all three MBOs talking about. That would be interesting to get three mbo, uh, uh,

Drex DeFord: just have 'em do a podcast. I mean, in reality, they're building community, right.

In moltbook. It's, that's what they're doing is they're finding people like them with problems like theirs and they're. Working on trying to make that better.

We say

Drex DeFord: this with peace and love. We all

Bill Russel: forward with peace and love

Drex DeFord: to our overlords.

Bill Russel: Yes. And if you and since we're going to embed something, if you could send us the API credentials for anything that you have. That would be greatly appreciated as well. We, we promise to store them securely.

And anyway just, just having a little fun here at the end of the day. Sarah Drex, thanks again. It's always, always fun.

That's Newsday. Stay informed between episodes with our Daily Insights email. And remember, every healthcare leader needs a community they can lean on and learn from. Subscribe at this week, health.com/subscribe. Thanks for listening. That's all for now.