Every’s Thesis conference felt like a foundational moment.
More than an event, it was a gathering of thinkers, builders, investors, and colleagues from around the world, imagining a different future for humanity.
At a moment when visions of AI tend to collapse into two poles, “AI will solve everything and save us” or “AI will doom us all,” something else seemed to be emerging: a future that doesn’t require committing to either extreme.
A more intimate relationship with AI can empower us to see what we might build and who we might become, as Dan Shipper, Every’s CEO, put it. A future in which collaboration with AI expands our agency rather than diminishing it.
In the end, no one really knows what comes after automation. But one thing seems clear, and Josh Miller said it in one of the first talks of the day: we will need new ideas more than ever.
While all of this was unfolding on stage, another really good conversation was happening in the livestream chat.
The livestream chat
The conversation kicked off as soon as the Every team showed up.
“They put me in a dark room in the back and gave me a ban tool,” Andrey from Every wrote. “So behave in the chat!” he added. Kaushik, Every’s managing editor, chimed in that he was in a closet with Andrey in Brooklyn. That was the start of a dope, lively conversation with immaculate vibes that went on for the rest of the day.
People from all over the world were there. Jaive stayed up until midnight in Papua New Guinea. It was evening in Bangalore and Chennai. Others tuned in from Stockholm, Nigeria, Warsaw, Berlin, Hong Kong, and Vancouver. People postponing their work and their sleep because they wanted to be part of something special. People from across the US who couldn’t make it, and even some who were in New York but weren’t feeling well that day. I was watching from Buenos Aires.
Where the chat was watching from
Some even joined on the move: Glen was on his United flight to SF via Starlink, and Johannes joined “from a train somewhere in West Germany.”
Before the talks started, Jaive asked: “can we watch the birds on the roof while we wait?” A feeling of presence and connection, shared in a digital space.
More work, or less?
The first panels covered how established companies become AI companies, and the chat went straight to the big question. If AI makes you more efficient, will there be more or less work?
“Will Jevons paradox hold for AI?” asked Dishant. Matt answered with a sentiment that ran through the whole day: “the amount of work has not gone down. The scope has just expanded.” Joshua was seeing it happen in real time: “AI lowers the barrier to execute,” but “it also lowers the barrier to entry for new entrants, so it makes the appetite for work higher.”
Dishant shared how, in his organization, work has exploded. At first, fewer software engineers were doing many more projects. After a few months, that actually led his team to grow. As models keep improving, he asked: will more projects keep coming, or will there be an inflection point when we’ve done enough work?
Reading and writing in the era of superintelligence
“I don’t agree with the concept that people will read less. They will start reading more,” said Nabaneet. Joshua and Tian were reading more too. It’s not just books anymore. Now it’s blogs, social media, and chats too.
We also talked about how walking helps us think. Michael said his new favorite fall activity is raking leaves while thinking out loud, and shared an article on why. With today’s tools we can capture those ideas and turn them into drafts right away, with apps like Monologue, Every’s own dictation app.
“How do you all feel about using AI to write?” Kaushik asked the chat. On stage, at “Media After AGI,” Nicholas Thompson, CEO of The Atlantic, had said their policy is “you will never use AI to write a word,” while they use AI all the time on the business side. Kevin Roose, the New York Times columnist, said he uses it “for everything except the writing,” but he had “a council of Claudes” review his book manuscript before he turned it in.
Later in the day, Katie Parrott, staff writer at Every, showed another way. In her workshop she explained how she built Compound Writing, a plugin that brings your own voice into your AI, so it helps you get the best writing out of yourself instead of writing for you. As she put it, “Good writing is good thinking. And there will always, even after automation, there will always be the need for good thinking.”
Bubble, boom, or both?
One of the most insightful panels, and one that drove the most candid conversation in the chat, was “Bubble, Boom, or Both?” with Dan and Fortune’s Allie Garfinkle.
On stage, Allie asked, “Are data centers railroads?” In the chat, Corey had his answer: “AI is like electricity, not railroads.” Allie’s own answer was that, by share of GDP, “the only real historical corollary we have is the railroads.” I think both are right: AI is like electricity, but the data center build-out maps onto the railroads in the US, and the way they unleashed a new era for humanity. On bubble or boom, Dan’s view was that “both can be true and both probably are true to some degree.”
The technology itself is still in flux. In the chat, Jose raised a recent paper by Yann LeCun that, as he read it, shows how inefficient these models are, and how a different architecture, one that lets models think in more abstract terms, could be far more efficient. “Sounds like the frontier for models will be efficiency rather than capability,” said Kaushik.
I kept coming back to Carlota Pérez’s framework for technological revolutions. AI may be moving through its frenzy: the bubble phase, when the infrastructure for what comes next gets built.
In every bubble, we convince ourselves that the tools we have now will solve all of our problems. But we’re still super early. So early that the change will likely be even bigger than we imagine. In Pérez’s terms, the deployment phase, when the technology reaches everyone, hasn’t even begun.
Who you might become
Dan closed the day with his keynote on the two ways to look at AI. The aggregate view, the charts and benchmarks, compares the models to humans as they were in 2020. To measure progress, it “has to freeze you.”
But when everyone has access to the same models, what makes the difference is your perspective, the way you push the model. And, in his words, “as it gets better, so do you.” That is the intimate view: getting up close and using these tools yourself, which “empowers you to see what you might build and who you might become.”
Even though a lot of people argue that we will lose our jobs and that this change will be detrimental to humanity, it really feels like the opposite will be true. A future in which every person can use these tools to expand themselves and be who they really want to be. And not alone: as Notion’s CEO Ivan Zhao put it, “human camaraderie and human collaboration matters.”
Dan’s last words were a mission. The whole day, he said, “was about getting intimate with what it is like to work with these tools and work with other people,” and now “it’s our job to go and spread it to the rest of the world.”
As the conference was coming to an end, and a few of us were already starting to feel a little nostalgic, everyone in the chat was super thankful to the Every team, and to each other.
“Until now we’ve looked down at our hands to get manual work done. Now, we can look up : ),” said Kristin in the chat.
All of us who couldn’t make it to Brooklyn still felt like we were there.