ESD
Hey, welcome back to the ESD newsletter.
Here’s what we’re covering today:
🤖 Inside the Instinct Frenzy
📈 ESD Trending: 5 Hot Stories Lately
👨🏻💻 Best Advice for Startups Before Product-Market Fit (According to Brian Armstrong, Founder of Coinbase)
Let’s get started!
🤖 Inside the Instinct Frenzy

Credit: Noah Shinn
A 23-year-old college dropout built an AI app that launched just months ago.
It's still invite-only. It makes zero revenue. Most people can't even access it. And investors are now in talks to value it at $10 billion.
This is the crazy story of Instinct.
The founder
His name is Noah Shinn, and in 2023 he walked out of Northeastern University to build AI agents full-time. That could sound reckless. It wasn't.
By the time he dropped out, he’d already made a name for himself among serious AI researchers. He was the first author of Reflexion, an influential paper on teaching AI to learn from its own mistakes. Then he joined Sierra, the AI company run by Salesforce's former co-CEO, as one of its earliest scientists.
So when Noah left to start his own company, the smartest people in AI started paying attention.
The product
What he built is called Instinct, and it works unlike any app you've used. There's nothing to open. No dashboard, no interface. You just text it or call it, like a friend who happens to handle everything.
And it actually does the work. Ask it to book a flight, clean out your inbox, cancel a subscription, or order groceries, and it goes and does it. One early user is reportedly planning their whole wedding through it.
That's the part that turned it into a phenomenon. People who got access wouldn't shut up about it, and invitations spread on their own. The waitlist swelled so fast the app started warning users it was running at full capacity.
The money
And it followed fast. Instinct had already raised around $350 million from top investors in a matter of months. Now Sequoia and Benchmark are in talks to lead its biggest round yet: $1 billion at a $10 billion valuation.

Which is where the story gets strange. Instinct doesn't make money, and Shinn has said he doesn't even want to charge people to use it. So most of that billion dollars wouldn't fund growth. It would buy computing power, just to keep the free users running.
To feel how strange that is, put it next to companies you know. Figma, the design tool millions of professionals pay for, makes over a billion dollars a year. Snapchat has nearly 500 million daily users and billions in revenue. Instinct is chasing the same tier with a hundred thousand invite-only users and nothing coming in.
And it got there almost overnight. In late August, the company was valued at $2.5 billion. Weeks later, it was in talks at four times that. For a product almost no one can actually use, that kind of leap is hard to explain with numbers.
The real bet
Because the numbers aren't what investors are buying. They're buying the founder, the FOMO, and a belief that this is how everyone will use AI in a few years.
Just a couple years ago, Noah Shinn was a student who walked away from his degree. Today, the most powerful investors in the world want to value his months-old app AT $10 billion, before it has earned a single cent.
So which is it? The AI bubble at its peak, or the moment agents finally start to work?
ESD Trending: 5 Hot Stories Lately
A London startup got $25M for beating humans at guessing the future
Mantic built an AI, entered it in the Metaculus Cup, and let it compete against actual forecasters. Then Radical led a $25M seed round. Balderton, M12, and Thinking Machines Lab piled in. The pitch writes itself: if your model predicts the future better than people, stop calling it research. Sell it.
Muse shops, books tables, fills out forms, and calls businesses. It launched in the US this month, added a Mac app, and hit #1 on the free chart. Pricing: free, $20, or $100. Consumer agents aren’t a slide deck anymore. People actually downloaded one.
Anthropic wants the biggest IPO ever - while Claude builds the next Claude
November is the target. Valuation chatter is now in $2 trillion territory. OpenAI is sitting out until 2027. Inside the building: Claude now leads ~26% of Anthropic’s R&D, up from ~0% in February, with tens of thousands of agents running. Also this week: please slow down, says the company flooring it.
Cognition is now worth $48 billion
That’s the company behind Devin, raised $2B and nearly doubled since May. Revenue is heading toward $900M. Customers include Nvidia, GE, Citi, and Mercedes. The job is no longer “help me code.” It’s “take the ticket.”
You remember the mess: viral agent, Meta agrees to buy it, Beijing kills the deal. That part’s done. The split wrapped in May, Meta’s stake was bought out at about $2B, and Manus is independent again. Now it’s raising around $500M at a $4B valuation and talking about a Hong Kong IPO.
Best Advice for Startups Before Product-Market Fit (According to Brian Armstrong, Founder of Coinbase)
“If you’re pre-product/market fit, the best advice that I have from that period is: action produces information. Just keep doing stuff.”
He borrows this idea from Paul Graham, who said:
“Startups are like sharks. If they stop swimming they die.”
Brian’s point is simple: even if you’re not sure what to do, do something.
“When you do it, it’ll produce some information. People liked it, they didn’t, etc.”
There were many times in the early days of Coinbase when Brian shipped something rather than debating it endlessly:
“And there were a couple of times where the minute I shipped it, I knew we built it wrong. But I now had an idea of what to do next. And I only would’ve had that idea by actually going through the exercise of building it.”
His favorite analogy:
“It’s like you’re at the base of a mountain that’s shrouded in fog. You’re looking up at the mountain and trying to figure out how to get up there, but you can only see three or four steps ahead because the fog is so thick.
So you have to just take steps into the unknown. And when you take three steps, another three steps will be revealed ahead of you.
And sometimes you’ll end up on some local maxima and have to retrace your steps. But most people in life don’t take the steps into the unknown because it’s scary.”
The real lesson: You don’t need to know the entire path. You just need to take the next step that gives you more information.
Follow Us on X
We're posting our best stuff on X first, before it hits anywhere else.
We just got the account going, so if you're around, give us a follow🙏🏻
That’s a wrap for today 🫡
Tell us what you think. We’ll read and reply to every email.
Until next time,
Komron & the ESD Team

