A week in AI is like a year in other industries. I hope these issues become your weekly source of AI information, inspiration, and ideas. If we haven’t met before, I’m Amanda Smith. I write about AI and the fascinating folks who are building in this brave new world.
Good morning. This newsletter publication takes you behind-the-scenes of AI companies at all stages, in all kinds of industries. But it’s not every week that we get to talk to the AI leaders building within established, well-known companies.
If you’re interested in reading how one of the most ambitious internet companies is building in AI, you’ll love this issue.
Let’s dive in.
This week in AI
AI labs are spinning up side implementation companies
They’re betting on this new category.
Australia imposes energy and water guardrails on data centers
In a landmark AI masterplan.
New York becomes first state to impose data center moratorium
The construction ban will apply to data centers that use 50 megawatts or more of power.
🤝 This edition is kindly brought to you by Atera
RMM, helpdesk, and AI that isn't just an add-on. It's embedded across every workflow. Query your fleet in plain language, generate scripts on demand, and get insights from your live environment without digging through reports.
No integrations to stitch together or AI layer bolted on after the fact. Just one platform that already knows your stack."
Jay Hack, ClickUp
Jay Hack was interested in math from a very young age. He walked by a lab that had a robot exhibition and approached a professor, asking if he could fetch him coffee and work in his lab. There’s a program in Ann Arbor that allows high school students to take classes at the University of Michigan and get credits, which Hack did for two years. This was in 2008, pre-deep learning and Hack cut his teeth in very classical methods in AI.
He went on to study computer science at Stamford. One of the seminal works, Gödel, Escher, Bach set his life course and Hack has since been
entranced by ideas around consciousness, cognition and computation.
Hack spent 18 months of his working career going into Fortune 500 companies, deploying machine systems on data they already had to make predictions. Interested in consumer tech, he co-built one of the first applications of computer vision to a consumer problem – helping shoppers find products that are appropriate to them.
“We built this massive data asset of videos of influencers talking about products as well as reviews of products. We applied computer vision and natural language processing on top of that, to refine that data to come up with a corpus of information that can be used to make informed recommendations for consumers.”
“We were like intelligent sample distributors because of the data we had and customers were able to give them products we thought they would convert to,” Hack said.
That company was bought by a private equity firm, in a deal that included Birchbox.
This was around the time when the GPT3 paper was announced, and it quickly became apparent to Hack that the incoming wave of language models was going to be code generation. He started building things with AI code and sharing a demo on Twitter – including being the first person to launch a text prompt to create a Figma document.
Hack spent the next three years building one of the first coding agent companies, Codegen.
“Codegen specailized in integrations with existing platforms. We thought the way to win was to meet the user where they were at, instead of building our own interface.” Codegen integrated directly into platforms like Clickup, Notion, Monday.com. They became the fastest growing and largest agent on ClickUp, which Codegen was later acquired by in December 2025. Hack followed as Head of AI, taking the core architecture behind Codegen to modernize ClickUp’s agent stack.
He saw how optimally positioned ClickUp was to take advantage of the forces at play as agents roll out over the economy. ClickUp is vertically and horizontally integrated, Hack said, unlike other platforms that don’t have the data or user engagement in one place – so they’re having to stitch together multiple platforms and are non-permissive in what data external agents can see.
It’s never been easier for horizontal platforms (like ClickUp) to add features that end up looking like vertical software platforms. Hack predicts the software economy will have far fewer isolated software companies.
The golden age of productivity
“Everybody is going to be a software engineer in the future. Every company three years from now is going to look like a tech company three years ago. A lot of the problems that the world has can be encoded in code,” Hack said.
“Taking it as an assumption that software production is basically free and you have access to a country of geniuses in a data center who can rip code for you, that opens you up to go off and pursue all of the other problems. People should take software for granted and go off and solve harder stuff in the physical world.”
In terms of his team at ClickUp, they’ll see new capabilities come out and ferociously ship them to provide a better experience to users. “I’ve shifted from somebody who’s just vibe coding myself all day to now mostly managing an organization of people and making sure we’re focused on building the right thing,” he said.
“It’s amazing to be launching products at a scale where we know people are going to immediately use something anytime you flip on a feature and being able to see consumer dynamics play out.”
ClickUp looks more like a start-up than a corporate bureaucracy, although they’re soon to cross $400 million in revenue.
“There's maybe between five and 10 teams in the world that actually have a shot at making general-purpose knowledgeable agents work at scale, and we're one of them. So, it's just going to be a sprint,” he added.
The biggest market segment
Hack believes companies need to start re-conceiving their customer base not as humans but rather as agents. Handing over the keys to agents is going to be a species level event.
People are going to be able to build businesses from scratch and operate zero-person companies. “We’re going to increasingly find ourselves in a position where we’re marketing products to agents as opposed to humans, because the buyers are going to be the agents and the business operators are going to be agents.”
Right now, an agent can perform a 15-hour task with a 50% hit rate, Hack explained.
He advised leaders to look at world models or physics-based foundation models that also work with text and find what the opportunities are there. In terms of company differentiators, Hack said data is the easiest one to point to.
“I’m not convinced that there’s actually going to be a moat besides existing customer lock-in and distribution at the application layer,” Hack concluded.
How's the depth of today's edition?
If one of these stories stuck with you, I’d love to hear which one.
Speak soon,
Amanda
P.S. If you want to get a founder feature about your own story, reply to this email. If you’d like to reach our newsletter audience (founders, creators, and marketers), click the button below.
If you’re new here, I’m over the moon you’ve joined us! To help me craft content that’s actually useful (and not just noise in your inbox), I’d love it if you took 1 minute to answer this quick survey below. Your insights help shape everything I write.
✨ Insane Media is more than one voice
Dive into our other newsletters - where psychology meets the creator economy, e-commerce marketing, and Human resources.
