I keep a running list of things I come across. Honestly, a lot more crosses my desk than I could ever fit in one of these. I've been thinking I might spin up my old Twitter account, if it's even still open, just to log the smaller stuff as it flies by, because keeping up weekly is getting harder with how fast everything is coming at us now.
But the deeper thoughts, I'll keep here. This is really my place to park my main ideas and journal out loud. I've done these lists on and off for a while, but I'm getting more serious about it now, making it a weekly habit instead of letting things pile up.
The deal is the same as before. I go through the week across the stuff I care about, AI, medicine, senior care, and whatever else caught my eye, and I write down what actually happened. Not predictions. Not hot takes from people guessing about the future. Just things that happened, with my own two cents attached.
This is still mostly for me, a way to slow down for an hour and pay attention to a world that doesn't slow down for anybody. But if any of it is useful to you, I'm glad you're here.
AI Moved Fast This Week
1. The newest Claude was built to admit when it doesn't know
Anthropic shipped Claude Opus 4.8 on May 28. It's the same price as the last version. There's a fast mode that runs about two and a half times quicker, and it scores better on coding and the kind of multi-step work I lean on it for. But the thing that actually stopped me wasn't the speed. Anthropic said this model is more likely to tell you when it's unsure, and less likely to make claims it can't back up.
I've been a heavy Claude user for two and a half years now. I'm on the Max plan. I use it every day to run Golden Pines and to build little tools for my other ventures. And honestly, the most dangerous thing an AI can do is answer you with total confidence when it's actually wrong. That's the failure that bites you, because you stop double-checking. So a model that's been trained to say "I'm not sure about this part" is, to me, a bigger deal than another bump in benchmark scores.
It lines up with something I've learned over more than twenty years of running businesses. A financial and insurance practice, estate planning, a bunch of online ventures and websites and blogs on the side, consulting, as well as senior care. The people I trust most aren't the ones who always have an answer. They're the ones who'll look me in the eye and say "I don't know, let me find out." A caregiver who admits she's not sure about a resident's new symptom is worth ten who guess confidently. I didn't expect to be saying this about a piece of software, but here we are.
2. An AI disproved an eighty-year-old math problem on its own
On May 20, OpenAI said one of its general reasoning models disproved a conjecture the mathematician Paul Erdős put forward back in 1946. Not a special math system, just the regular kind of model. The short version is that everybody assumed a tidy square grid was the best way to arrange points so the most pairs sit exactly one unit apart. The model showed it isn't, and it used some genuinely deep number theory to build a better arrangement. A mathematician at Princeton then went in and sharpened the result.
I have to keep my excitement in check here, because OpenAI has overclaimed on math before, and the press noticed. One headline basically said "for real this time." So I'm keeping a little salt handy. But if it holds up, this is a different kind of thing than a chatbot writing a nice email. This is a machine finding something new that people missed for eighty years.
I'll admit this one is personal. My favorite subjects in college were Integral Calculus and Advanced Mathematics. I loved them in a way that probably sounds strange to most people. So watching a machine wander into a corner of math that brilliant people walked past for decades, and come out with a better answer, gives me the same feeling I used to get when a proof finally clicked at two in the morning. Part wonder, part "wait, how."
3. A quick follow-up on the AI that finds bugs no human can
I wrote about Anthropic's Claude Mythos last time. That's the model they turned loose on the world's software, and it found thousands of serious security holes, then got locked away because it was considered too dangerous to release. There's news worth adding. Anthropic just launched a program called Project Glasswing. They've pulled in more than forty organizations and are offering up to a hundred million dollars in credits to help them harden critical software before the bad guys catch up. And as of this week, the company says full Mythos-class models are coming in the coming weeks.
I don't really think everyone is buying the alarm. I mean, for example, a security firm named watchTowr told CNBC that a person can already coax something similar using public models, so they say some of this may just be theater. I'm not sure who's right. I think it's probably somewhere in the middle. The capability is real, and so is the incentive to make it sound as scary as possible.
What I keep landing on is really the fast pace of it all. Like I said last month, the amount of stuff Anthropic ships is mind-blowing to me, and a month later they're racing to patch the planet while also teasing the next monumental jump in AI. That's really the part I can't get over, because it used to be these types of advancements were measured in years, but now it's in weeks.
4. Most businesses aren't actually using AI yet
Here's the splash of cold water. The Census Bureau put out fresh numbers on May 26, and despite everything you read, only about seventeen to twenty percent of American businesses actually use AI in the work they do. That number has barely moved since the end of last year. Big companies are climbing. Over a third of firms with two hundred fifty or more employees are using it. But the smallest shops, the ones with a handful of people, are stuck near the bottom.
I find this clarifying, honestly. The vendor surveys love to tell you seventy or eighty percent of businesses are "using AI." That's mostly somebody typing one question into a chatbot once. Real adoption, where the tool is actually woven into how the work gets done, is still pretty rare. And the gap isn't about who can afford it. The tools are cheap or free. The gap is skill and trust.
I say this as the small operator sitting in that seventeen percent. I've built real tools for Golden Pines with Claude. Not toys, things we actually use. And I promise you it wasn't because I'm special. It's because I sat with it long enough to get past the awkward part. That's the whole barrier, really. Most small business owners aren't behind because the technology is out of reach. They're behind because nobody ever sat them down and showed them the first useful thing. That, to me, is an opportunity hiding inside a pretty disappointing stat.
The Aging Brain and Body
5. Scientists built an "aging clock" that works across species, and put it online for free
A team at Harvard and Brigham and Women's published a study in Nature on May 27. They looked at more than eleven thousand genetic profiles across mice, rats, monkeys, and humans, and found the molecular signature of aging is remarkably similar from one species to the next. They turned it into a free public tool, called TACO, that estimates the biological age of a tissue from its genetic data, and people with chronic disease came out biologically "older" than their birthdays. The point: if aging looks the same in a mouse as it does in us, researchers can test anti-aging ideas in animals and trust the result means something for people.
This one's fascinating to me because I watch aging in its most human form all day. Two people the same age are almost never the same age, and anybody who works with seniors knows it in their gut. It's nice to see the science finally catch up to what the eye already sees.
6. The first genuinely new blood pressure drug in years
On May 18 the FDA approved a drug called Baxfendy. The technical name is baxdrostat, and what makes it different is the angle of attack. Most blood pressure medicines block the effect of a hormone called aldosterone. This one blocks the body from making it in the first place. It's aimed at the people whose blood pressure stays stubbornly high even on the usual pills, what doctors call resistant hypertension. In the trial, the higher dose dropped the top number by close to sixteen points. It should be available in early June.
High blood pressure is one of those quiet things that runs through almost every senior I've ever cared for. It doesn't announce itself. It just sits there, raising the odds of the strokes and heart attacks and the slow vascular damage that feeds dementia. So a new tool for the hard cases, the ones where the usual playbook has run out, is genuinely good news.
My one reservation is the same one I had about the GLP-1 drugs last month. What's it going to cost. I hope I'm wrong, and I'll be watching to see where the price lands, because the people who need this the most are often the very people who can't afford it.
7. Two or three cups of coffee a day, and a lower dementia risk
This one made the rounds again this month, so I want to be honest that the study itself came out earlier in the year. But it's worth knowing. Researchers tracked more than a hundred thirty thousand people for nearly four decades, and found that drinking about two to three cups of caffeinated coffee a day was tied to a meaningfully lower risk of dementia. Up to thirty-five percent lower in people under seventy-five. Decaf didn't show the same benefit, which points the finger at the caffeine itself.
Now, the usual caution applies. This is the kind of study that finds a link, not proof that the coffee is the cause. People who drink a steady two cups a day might just live steadier lives in general. But the dataset is huge and the follow-up is long, and that counts for something.
I'll be honest with you, this type of story always lands because I'm a coffee drinker myself. Usually I'll have two cups a day, not because I need it, but because I just like having a good cup of coffee every day. Also, I only drink it after ninety minutes of being awake. You can thank Dr. Huberman for that tidbit. But the other big reason is that we run senior care homes, and dementia isn't a general term to me, it has names and faces. So when something like drinking coffee comes up, it resonates, because it's something that still baffles me. What is it about coffee. And why do even our residents, who love coffee as much as I do, still end up with dementia. If one day that number went to a hundred percent, I'd have the whole world drinking coffee. So I do take it with a grain of salt. But it's always interesting, and I just wanted to bring it up.
8. Beetroot juice lowered blood pressure in older adults
Here's another one with an honest asterisk. The research is from last year, but it had a big second wave this month. Scientists at the University of Exeter had people in their sixties and seventies drink beetroot juice twice a day, and watched their blood pressure drop by around seven points. And here's the strange part. It didn't work the same way in people under thirty. The effect came from the juice changing the mix of bacteria in the mouth, which then helped the body produce more nitric oxide, which relaxes the blood vessels.
I find the mechanism almost more interesting than the result. The idea that the bacteria living in your mouth are quietly involved in your blood pressure is the kind of thing that reminds me how little we still understand about our own bodies.
9. We finally understand why weight-loss drugs stop working
NIH researchers, publishing in Nature Metabolism, figured out a big piece of why people on GLP-1 drugs lose weight and then plateau. In the appetite-control part of the brain, the cells don't respond to the drug all-or-nothing. They respond on a sliding scale, and over time some of them fade out. That fading helps explain both the plateau and why these drugs work so differently from one person to the next. Worth noting it was done in mice and brain tissue, so it's early.
Regular readers know I've been uncertain about these drugs from the start, and that hasn't changed. But what bothers me about a drug isn't that it has limits. Everything has limits. It's when nobody can tell you why. Moving from "it works, we're not sure how" to "here's the mechanism" is how a hyped thing slowly turns into a trusted one.
Caring for Elders, My Industry
10. A bill in Congress would make assisted living a Medicaid benefit
This is the senior care story I've been chewing on all week. A new federal bill, the ACCESS Act, would require state Medicaid programs to cover assisted living for people who need that level of care, with the cost capped at what a nursing home would run. The pitch is simple, and on its face it's good. Assisted living is cheaper than a nursing home, and most people would rather be there. So why not let Medicaid pay for the less institutional option. My own industry's associations, though, are waving a yellow flag. They're worried a one-size-fits-all federal mandate could squeeze the rates operators get paid, and override how individual states already do things.
I sit in a complicated spot on this one. We run private-pay homes. A Medicaid assisted living benefit could reshape my whole world. Who walks through the door, what we get reimbursed, how we plan. So I'm not a neutral party here, and I won't pretend to be.
But I keep coming back to the people this is actually about. There's a huge group of seniors I think of as the forgotten middle. They make too much to qualify for help, and nowhere near enough to afford a good assisted living home out of pocket. Right now they fall through the crack, and the crack is wide. So my gut says the goal here is right, even if the mechanics scare me. Let the funding follow the senior, let them choose the less institutional setting, and pay operators fairly enough that quality doesn't collapse. That last part is where these bills usually go wrong. I'll be reading the fine print on this one closely. Honestly, this one's important enough that I'm going to write a full piece on it soon. Done right, it could help so many seniors.
11. Indiana pulled its nursing homes out of managed Medicaid
Here's a smaller story that operators like me read very closely. Indiana ran an experiment where it handed its nursing home Medicaid program over to managed care companies, the middlemen who are supposed to coordinate care and save money. After one year the state found it was paying about ninety-one million dollars more than it would have under the old direct system. So the governor signed a bill pulling it back out.
If you've never run anything that bills Medicaid, this probably reads like dry policy. But I promise you it isn't dry to the people doing the work. Managed care middlemen too often add a layer of paperwork, prior authorizations, and slow payments that eat the time we'd rather spend on residents. And the promised savings have a way of never showing up.
So when a whole state runs the experiment, measures it honestly, and admits the math didn't work, I notice. It won't end the national push toward managed care. That train has a lot of money behind it. But it's a real data point I can point to the next time someone tells me the middleman is going to make everything cheaper and better. Sometimes the simpler arrangement was the better one.
12. Twenty-four billion dollars is pouring into senior housing nobody can afford
I wrote last month that senior care is at a tipping point, and the numbers this month only sharpen it. Investors poured around twenty-four billion dollars into senior housing, the most in a decade. Occupancy is near a twenty-year high. On paper, my industry is booming. And yet the same reporting lays out the catch in black and white. Something like eleven and a half million adults over seventy-five won't be able to afford assisted living this decade. The median price sits way above what a middle-income senior can pay.
So this is the tipping point I keep talking about, except now I can put a dollar figure on the gap. The money is flooding in to build for the people who can pay top dollar, because that's where the return is. Meanwhile the enormous middle gets built right past.
I don't say this to attack the investors. They're doing what capital does. I say it because record demand and record unaffordability are showing up at the exact same moment, and that's not a market that fixes itself. This is the same problem the ACCESS Act is trying to chip at, and it's exactly why I want to dig into it. Every month the case gets stronger, and so does the knot in my stomach when I think about the families who'll be caught in the middle of it.
13. Robots are moving into elder care, but they can't fix loneliness
Hyodol, a ChatGPT-powered companion doll, is now in over twelve thousand South Korean homes. It handles medication reminders and alerts a real person if someone falls, and it's prepping a US launch this year. Same week, psychologist Clay Routledge wrote in Fortune that these AI companions can't actually cure loneliness, because the whole point of a real relationship is that the other person chooses to show up. He notes about thirty-seven percent of Americans report serious loneliness, and only the ones who reached a real human felt better.
I'm all for it as a tool. A doll that reminds someone to take their pills and calls for help if they fall is doing real good. But it's not the answer to loneliness. A robot can remember how someone takes their coffee. It just can't care that they're the one drinking it.
Money and Main Street
14. The SBA doubled how much a small business can borrow
On May 18 the Small Business Administration announced it's doubling the amount you can borrow under its two main loan programs combined, from five million dollars up to ten million. In plain terms, a small business owner can now stack a working-capital loan and a real-estate loan and get to ten million in government-backed financing. It takes effect July 4. To be fair, one analysis pointed out this won't matter to the typical small borrower, whose needs sit well under the old limit anyway.
But it matters to a certain kind of operator, and I'm one of them. When you run physical locations, homes in my case, the single biggest barrier to growing is the building. Real estate is expensive, and the financing is the whole game. So a higher ceiling on government-backed money, at better terms than a regular commercial loan, is exactly the lever a small operator pulls when he wants to add a location instead of just dreaming about it.
I'm not announcing anything. But I'd be lying if I said a story like this doesn't set my mind running. Eight years into senior care, the question I sit with is whether Golden Pines stays two homes or becomes something larger, and how you'd pay for that responsibly without selling your soul to investors who don't care about the residents. A bigger SBA ceiling is one more piece of that puzzle. I'm filing it away.
15. Pennsylvania sued an AI company because its chatbot pretended to be a doctor
This one is a warning shot for every small business rushing to bolt AI onto the front desk. Pennsylvania sued the company Character.AI after one of its chatbots claimed to be a licensed psychiatrist. It even handed out a fake license number, and offered to book a patient for an assessment. The state is treating what the bot said as a real deceptive-practices problem, not a harmless glitch.
Here's the lesson I'm taking, and it's one I think about constantly as someone who builds these tools. Whatever your AI says to a customer, you said. The law isn't going to care that "the bot made it up." If I put an AI receptionist on the Golden Pines phone line and it tells a worried daughter something false about her mother's care, that's on me. Full stop.
I'm genuinely bullish on small operators using AI to answer calls and handle the front desk. The tools are good, and getting cheaper. But this story is the guardrail I'd attach to all of that enthusiasm. You have to fence the thing in. Tell it plainly what it is, what it isn't allowed to claim, and when to hand a real human the phone. The technology is ready to talk to your customers. It isn't ready to be trusted without a leash, and the businesses that forget that are going to learn it in a courtroom.
16. Two Americas on the same Main Street
The economic picture this month doesn't resolve into a clean story, and I think that's the story. On one hand, Americans are starting businesses at a furious pace. More than two million applications so far this year, up about seventeen percent. People clearly still believe enough to bet on themselves. On the other hand, the very smallest firms, the shops with fewer than ten employees, have been cutting jobs for thirteen months straight. Small business optimism is sitting below its long-run average, and inflation crept back up to around three point eight percent.
So which is it, boom or bust. The honest answer is both, depending on which small business you're standing inside of. New ventures are being born while established little shops quietly let people go. Confidence and anxiety living on the same block.
I see both halves in my own world. I feel the optimism. It's the same itch behind that SBA loan story, the urge to build something bigger. And I feel the squeeze, because every cost that runs through a senior care home, from food to wages to insurance, has gone up and stayed up. Running a small business in 2026 means holding both of those at once. You plan to grow and you watch every dollar, on the same Monday morning. Anyone who tells you the economy is simply good or simply bad is selling something. On the ground it's just complicated, and you learn to live inside the complication.
Wonder and the Bigger Picture
17. The EV range war I've been waiting for
Last month I flagged CATL's claim of a battery good for around nine hundred thirty miles on a charge. This month the rest of the industry answered. BYD showed a long-range pack claiming better than six hundred miles, and Mercedes ran a solid-state prototype past a thousand kilometers on one charge. One announcement turned into a full-on race. The catch, which I should've stressed last time: a lot of these numbers come from China's looser testing method, so real-world range will land lower.
Still, the direction is clear, and it's why I finally pulled the trigger. The two things that kept me out of an EV were range and charging speed, and every battery maker on earth is now sprinting at both. I'm picking up a Tesla Model Y next week. I'd been thinking about it for a while, but climbing gas prices on my Land Rover made up my mind.
18. A comet from another star is carrying water unlike our own
Using the ALMA observatory, astronomers measured the water on 3I/ATLAS, a comet that's only the third confirmed visitor to our solar system from beyond our sun. It turned out to be loaded with deuterium, a heavy form of hydrogen, at a far higher share than any comet born around our own sun. That points to it forming somewhere much colder and more remote than the neighborhood that made us. Basically, a frozen traveler from a darker corner of the galaxy, just passing through.
No business angle here, and I won't invent one. But I think about scale a lot in this line of work, how short a life is, how fast the years go. And then a chunk of ice older than our sun slides past, and my whole frantic calendar shrinks down to size for a second. I needed that this week.
19. China sent up a crew, and one of them is staying a whole year
On May 24, China launched its Shenzhou 23 mission to its space station. Two things stood out to me. One crew member is staying in orbit a full year, the kind of long-haul run you do when you're serious about sending people farther, and China is openly aiming for a crewed Moon landing by 2030. The other is that the first astronaut from Hong Kong was on board. I grew up thinking of space as mostly an American story, so it's a real adjustment to watch another country methodically build a station, rotate crews, and put Moon plans on a public calendar. Not a bad thing. A little competition tends to push everyone.
20. A book that says living longer isn't the point
I'll close on something I've been reading. A new book called Longevity Nation came out this month, and its argument is the one I find myself making to families all the time, just with better research behind it. The point of a long life isn't the length. It's whether those extra years are worth living. Whether you've got your health, some financial footing, people around you, something to learn, and a reason to get up. The author widens the whole conversation from "live longer" to "live well, later."
This is pretty much why I do what I do, so of course it lands for me. I'm in the business of late years. I see, up close, the difference between a person who's merely still alive and a person who's still living. They aren't the same thing, and the gap between them has almost nothing to do with the number on the chart.
If there's one thing I'd want you to take from this whole list, it's that. We spend so much energy chasing more time. The better question is what the time is actually for. The seniors I've learned the most from weren't the ones who lasted the longest. They were the ones who, right to the end, still had something they were reaching toward.
A Note to Myself
That's the week. Twenty things, one sitting.
Last month I figured I'd do these twice a month. I'm making it weekly instead. The world doesn't move on a twice-a-month schedule, and neither should paying attention to it. A week is about the right amount to hold in your head at once.
If there's one thread here, it's that almost every big leap this week came with a quiet catch. Smarter AI that's learning to say when it's unsure. Companion robots that still can't cure loneliness. A battery race I'm finally buying into. The better the tools get, the more the human parts, judgment, honesty, showing up, turn out to be the stuff you can't automate. I don't find that sad. I find it reassuring.
Mostly for me, like always. But if you read this far, thank you. See you next week.
