Assisted Living
The agent boom is coming. Ready or not.
Hello from Meanwhile Brewing in Austin. Regulars, welcome back. To all of the newcomers this week, thanks for joining us.
It happened: an inch of snow fell in Austin, and we managed to keep the power on. Ted didn’t have to run away to Cancun, I didn’t have to wear a parka inside my house. You have to take those little wins, especially because winter is pretty to look at but sucks to live through and I am ready for it to be over. I am sure everyone in New York and everywhere in the country that dipped into single digits and below this week are overwhelmed with sympathy for my cause. Fortunately, it’s warmer everywhere now (though still subzero in the frost belt). Even better news: we are not here to talk about the weather.
It felt like a month’s worth of stuff packed into a compact car of a week. The kind that starts with $500 billion public/private AI partnerships named after Kurt Russell movies1 and ends with groundbreaking new DeepSeek AI model out of China that make Silicon Valley’s AI leaders question everything. Netflix announced its biggest quarterly subscriber gain in history, then raised prices by 15%, and everyone said ‘ooooh’ then gave a huge ¯\_(ツ)_/¯ which is at least better than the Oscars, whose 2025 nominations seemed to be met with the worst mix of indifference and a sort of derision in unison towards its most-nominated film that I can’t remember, even when Crash or Green Book took home the grand prize.
That’s just scratching the surface, without even digging into a full-ass inauguration and all of the trimmings; the stories and memes and Nazi salutes and Zuck staring straight into the sun. Not to mention the subsequent flurry of proclamations–some perhaps constitutional and others much less so–out of the newly-regilded Oval Office. The year’s first act opens with the subtlety of Emilia Pérez, the pacing of The Substance, and the foreboding techno-overlords of Dune.
We’ll bite off what we can today, but before we get started…
For my friends in Southern California: I will be in Los Angeles from February 4-7. I am headed to town for the BRXND.ai event in Hollywood, hosted by friend of the newsletter Noah.
The consensus on the two NY events seems to be that it was one of the best places you could go to have more practical, well-informed discussions about AI’s future in business and marketing. The type of conference where everyone comes in admitting what they don’t know, not spending the entire event trying to convince you they know more than they do. I loved this comment from an attendee:
Humility is the last thing you'd expect to find at a tech conference, but the emergence of generative AI demands it, and the humble, hopeful, even playful vibe at the BRXND conference felt like the exact right way to approach the brave-new-worldness of it all. To be at the first of what should be at least a semi-annual meetup was super energizing.
If you’re out in LA on February 7, you should join me there. Noah tells me they still have a few tickets left, get yours here:
And if you’re in L.A. that week but can’t make it, drop me a line, let’s make it happen.
Oh, one more thing before we get started: everyone please wish a happy 75th birthday to my incredible Mom, who still has a lot more Instagram followers than me.
Now let’s get down to it:
Secret Agent Land
On January 5, OpenAI founder and CEO Sam Altman made a blog post that rendered a pretty good picture of what to expect from his company this year:
We are now confident we know how to build AGI as we have traditionally understood it. We believe that, in 2025, we may see the first AI agents “join the workforce” and materially change the output of companies. We continue to believe that iteratively putting great tools in the hands of people leads to great, broadly-distributed outcomes.
AGI (artificial general intelligence) has been hyped for some time, it is for many the holy grail of AI and machine learning (and possibly the end of humanity, if you ask the guys building it), where AI “possesses human-like intelligence and can perform any intellectual task that a human can. It is capable of learning, reasoning, and adapting to new situations.”
Right now all the buzz is about “agents” as a major step in that direction.
In AI terms, agents are essentially automated workers who can carry out multi-step tasks, including any required reasoning, independent of human oversight. This analogy might be the easiest way to understand it:
This week, OpenAI formally introduced their first big agentic offering: Operator.
The early reviews from tech reporters were pretty tepid (or at least that $200/mo is very pricey for what it is today). It’s to be expected that this would be rough and expensive, and to be honest it may stay that way for awhile. But if you look at the amount being invested in making agents happen, and the overall velocity of innovation–the aforementioned DeepSeek emerged out of China this week, and people have already figured out to use it and open-source software to make your own agents for almost nothing–it’s reasonable to assume somewhere in the near future, an entire generation of agents will be unleashed on the internet and the rest of the connected world.
There are fundamental questions to ask: What do we do with that newly-minted free time? What if they take all of our jobs in the process? If we were worried about the TikTok algorithm, how will we feel about Chinese AI acting on our behalf across the internet? How many gallons of precious water will we use just so Skye can find just the right place for her sister’s bachelorette in Joshua Tree without having to search Airbnb herself?
Indulge me for a second and set those aside, along with your understandable skepticism about early agent examples like Operator. Let’s assume by this time next year that these agents are, like many of the early iterations of apps like Facebook or Uber that transformed our lives, good enough.
It’s 2026 and agents can carry out tasks like managing our groceries and calendars and daily lives, taking away what had been a lot of laborious2 tasks. They take off, millions of people have multiple agents doing various tasks for them. Which leads to my trillion-dollar question:
How does our world have to adapt and change, when everyone has one or more virtual counterparts doing their bidding every second of every day?
If you watch the early demos of Operator and other agents, you’ll notice that they aren’t all neatly tapping into systems intended for use by armies of automated bots. They’re pretending to be actual humans, clicking around, bypassing human limitations that we literally built the entire internet around.
What do the next few years look like if agents are brute forcing their way to automation, each of using one or more agents that dramatically increase the population of users on the internet? How do those of us who work on the corporate side, who are responsible for marketing and commerce and user experience, have to reinvent the way we work and what we make?
These are just a few of the thought exercises I’m using to develop the mental pliability needed for what lies ahead:
What do I have to do to get a table, or concert tickets, in a world where everyone has an army of agents vying to do the same thing? Does everything just go to the highest bidder and/or agents with the most compute power?
What does customer service look like in a world where anyone can spin up agents to bombard your call centers and help chats?
When making online purchases on our behalf, if an agent makes a mistake, like ordering the wrong product3, accidentally booking the wrong flight, one that can’t be changed? Will companies provide the same leeway they do to individuals?
Similarly: if my agent gets swindled and spends my money on a scam, are the bank and credit card company going to provide me the same protections?
If my agent is handling the transactions and communications, what is my “customer experience” and relationship with any brand, beyond their ads and the product itself?
If AI is indeed the future (and as important of one as we’d be led to believe), then I’d argue it’s vital that we don’t go forward blindly but rather ask these questions upfront so they can help inform where we ought to go next.
We also need to embark on the next chapter with the humility4 to acknowledge that many of the use cases proposed currently are really just bandages to address the things we designed and built–then, in many cases, underfunded and stopped improving–in the last two waves of innovation (desktop/internet then mobile/cloud).
We went through a period of great invention and brought to market multiple world-changing technologies in the span of a decade. The spectrum of possibility seemed limitless. What happened? The relationship between human-consumer and brand-enterprise grew increasingly antagonistic in the last two decades, thanks in large part to intermediating a lot of human interactions through those magical digital interfaces that were supposed to be a portal to a better world.
Are we going to let AI become one more layer of abstraction, replacing frustrating digital interfaces with smug virtual representatives who make it just as hard (if not harder) to get what you need?
Our esteemed leaders in Silicon Valley love to talk poker5 and our options here are similar: We can double down, ignore the mistakes of previous innovation cycles, and see if we can bluff our way to better. Or we can use AI as our opportunity to explore questions like those above, and truly improve our hand6.
The Best Snyder Cut
Daniel Snyder owned the NFL’s Washington Commanders from 1999 to 2023, and is widely considered to be one of, if not the worst owner in American professional sports’ modern era. If you’re not familiar with his work, the team’s hometown Washington Post has a good chronology of Snyder’s path to being forced into selling the team by the rest of the NFL’s owners.
This week, the team Snyder owned as recently as 18 months ago, who already have as many playoff wins in one season as the team did during all 24 years of Snyder’s time at the top, will play in the NFC Championship Game with a trip to the Super Bowl on the line. It has to be fairly embarrassing to watch how quickly the team got good after you disappeared.
Enter ESPN:
Yesterday, two of ESPN’s most senior and best connected reporters, Seth Wickersham and Don Van Natta Jr., dropped a 3,000+ word box of Morton’s kosher salt in Snyder’s self-inflicted wounds with a damning and at times humiliating look at the former owner’s final chapter and epilogue with the Commanders, and the seething animosity he’s left with watching his former team and their breakup bod dancing all up in the NFL Playoffs club.
As bad as all of that is, this story about the final handover of the team paints the worst picture of all:
Suddenly, the sale's closing -- a supposed formality -- turned into an eleventh-hour drama, multiple sources with direct knowledge told ESPN. Snyder threatened to kill the deal by refusing to share his bank information, preventing Harris from wiring him the money. At 1 a.m. on July 21, Snyder and his wife were fielding phone calls from various executives and confidants, urging him to do what he'd pledged and let go of the team.
"I don't want to do this," Snyder told a confidant.
A rally celebrating Harris' ownership group was scheduled for later that day at the since-renamed FedEx Field.
But as 1 a.m. became 2 a.m., Snyder was refusing to hand over the stadium keys.
"I don't care!" Snyder said, according to sources with direct knowledge of what transpired in those hours. "It would be trespassing if anyone goes there. It's still mine!"
Hilariously, steeping even more manure in the back of Snyder’s Maybach: this isn’t even his only recent source of very public embarrassment. He also helped finance the Trump biopic The Apprentice7, thinking it would be a “flattering portrayal” of the 45th and 47th POTUS’s early days in New York City, then tried unsuccessfully to block the movie from being released.
After selling the team for a record $6 billion he’s richer than ever. That money may buy him many things, but it won’t erase the long list of public failings as an owner (and, by all accounts, a person) during his time in Washington. On the field, his story will center around his crippling inability to find a quarterback after 27 tries.
His ongoing search for dignity may prove even more elusive.
Elsewhere
Deschutes Brewery makes big bet on NA beer (Axios)
Sweden is Building the World’s Largest City Made Entirely From Timber (Time)
Taylor Swift Triggered a Tsunami of Sports Bets. Then She Stopped Turning Up (Wired)
Men In Blazers Raises $15M From Media Heavyweights as It Preps World Cup Push (Hollywood Reporter)
Shout-out to friend of the newsletter Ricky, who is working his magic over there at Maximum Effort.
The biggest branding trends coming in 2025 (Fast Company)
Why Are Airlines Struggling to Keep Up With Soaring Demand? (Skift)
Cars That Time Forgot: The Berkeley Sports (Hagerty)
Project Stargate, though Project Overboard might well prove to be apt.
I use that very lightly, considering we’re talking about a series of clicks.
Before you say “but it won’t get it wrong.” It’s ordering things like clothing and food online. You really think there’s an A.I. coming soon that can make the call on which of the two jean cuts to go with based on a bunch of random size chart info and user reviews? A.I. can do many things but it won’t tell you that those pants will make your vagina look absurdly long.
Yes, I know we’re more likely to get Elon and Bezos to hug it out than we are to expect any of our leaders to summon even an iota of humility, but maybe we can carry the weight where their HGH and ketamine-charged shoulders cannot.
I assume they play too, but they may have agents for that now.
We can also point to all of AI’s failings and fold, sitting on the side and letting others control the outcome. That feels like the worst of all outcomes, smart people just walking away from a discussion this important.
Yes, he financed a movie featuring Jeremy Strong, who depicted Kendall Roy who sounds eerily like Daniel Snyder at times, the irony is not lost on over here.
Thanks for the shout out! Sunday Bunch is always entertaining and educational for this sometimes tech-challenged MOM.