I've been looking at AI product launches on X for 2 years.
And now my eyes just glaze over them. I can't tell them apart anymore.
AI startups have converged on the same fonts, the same gradients, the same launch arc, the same "AI for X" tagline, launch videos made by the same 5 people who make all X launch videos. The category aesthetic has become so unified that the brand can no longer do the one thing it's supposed to do: tell two companies apart.
The brand is built aesthetics-backward.
The investment goes entirely into the surface: the logo, the launch video, the typography, etc., that announce this is an AI company.
In a post-ChatGPT world, this is a fatal mistake.
Aesthetic is now free. A 1-hour Claude Code session recreates the whole stack: fonts, palette, layout, copy tone, the product demo, even the launch tweet. What used to differentiate a company is now table stakes. Everyone has it, which means no one has it.
So what should AI founders compete on?
Opinions
Real ones. Held in public. Funneling into decisions the company actually made.
Opinions are the most upstream component of a brand in 2026. Product, pricing, distribution, hiring: all downstream of what the company believes. The visual layer can be redesigned without killing the brand. The opinion can't.
What does an opinion look like in practice? Two examples from AI-native companies nailing it.
Granola: Every other AI notetaker built its growth loop on the same trick. A bot joins your Zoom call as a fake human, transcribes the meeting, then summarizes it. Every meeting the bot joins is a free demo to everyone on the call. The bot is the distribution engine.
Granola did it differently.
Chris Pedregal, Granola's CEO, sums up his opinion in his Ness Labs interview: “People hate these bots – they're awkward, they take up space in your Zoom call, they're overly invasive: recording audio and video.”
The opinion shows up in three concrete product choices.
- The product runs on-device. No stored audio.
- The interface is designed to look like a notepad, not like a meeting recorder. The user can type their own notes, and then the AI augments them.
- Early on, Granola decided to run frontier models instead of older ones. Pedregal said, “As a small startup with fewer users, we can use cutting-edge models that would be financially impossible for big companies to deploy at scale.”
3 years later: $1.5 billion Series C valuation and 400% ARR growth.
Clay: Most B2B SaaS companies treat sales as a numbers game and tooling as a boring system of record (Salesforce, ZoomInfo, Apollo): a database you query.
Clay disagrees. Kareem Amin's line, repeated across podcasts: Sales and go-to-market is a creative act.
If sales is creative, then the product, the team, and the brand all have to be creative too.
Clay built its brand intentionally: a full-time claymation artist hired before they had revenue; a 5-person brand team at 18 employees; SF billboards during INBOUND 2025 that read 'Every artist has their medium. Go-to-market has Clay’; over $1M spent on billboards in a single year.
Varun Anand, co-founder of Clay, remarked in his First Round Review interview, “When no one else is investing in something and you invest in something. That is a way to have alpha and that is a way to really stand out in a meaningful way.”
6 years to $1M ARR. 2 years to $100M.
Granola and Clay couldn't be more different as companies. But they are built on the same move: opinion-forward.
Cheap vs Expensive
But opinions on their own won't last.
Every AI company now publishes a manifesto. Every founder posts values on LinkedIn. Cheap opinions are everywhere.
An opinion is valuable only if it’s an expensive opinion — it's worth betting the entire company on.
It costs the company something. It walks away from revenue, customers, or markets that don’t fit with their opinion.
The companies that build enduring brands have expensive opinions. Two textbook examples.
Apple. Apple's brand is privacy, positioned explicitly against Android. And they dig their heels in. More than once.
In February 2016, the FBI demanded that Apple build a custom version of iOS to unlock the San Bernardino shooter's iPhone. Tim Cook published an open letter refusing: "The government is asking Apple to hack our own users and undermine decades of security advancements that protect our customers... We can find no precedent for an American company being forced to expose its customers to a greater risk of attack." Apple took the case to court and absorbed the political cost of being publicly seen as defending a terrorist's phone.
5 years later, App Tracking Transparency in iOS 14.5. An estimated $10 billion a year in foregone ad revenue for Apple. $36 billion in lost ad revenue for Meta in 2022 alone.
While anyone can write "we believe in privacy," only Apple has the receipts.
Anthropic. In July 2025, Anthropic signed a $200 million Department of Defense contract with two restrictions: no mass surveillance of Americans, no fully autonomous weapons.
In February 2026, the Pentagon demanded that those restrictions be removed.
Anthropic refused.
Three days later, President Trump ordered every federal agency to stop using Anthropic's technology. The Pentagon designated Anthropic a "supply chain risk to national security."
The cost is concrete: $200 million in immediate contract value, long cycles of litigation, and a federal government that doesn’t like you. But Dario Amodei held the line.
In his CBS News interview, he said, “The red lines we have drawn, we drew because we believe that crossing those red lines is contrary to American values. We're not gonna move on those red lines.”
That is what an expensive opinion looks like in 2026.
Controversial ≠ Expensive
One last thing.
A controversial marketing campaign is not the same as an expensive opinion.
Cluely launched in April 2025 with the tagline "Cheat on Everything." The brand was loud, controversial, and unmissable.
In June 2025, Cluely raised a $15 million Series A from a16z. The TechCrunch headline announcing the round read: "Cluely, a startup that helps cheat on everything, raises $15M from a16z."
Five months later, by November 2025, Cluely had repositioned as a standard AI meeting tool competing with Otter and Fireflies. The "cheat on everything" framing was gone. The product remained.
Seven months. That's how long the controversial brand lasted.
It's a simple test, really. Would the brand still hold the position if it cost them customers instead of winning them?
That is what separates a cheap opinion from an expensive one. Worth running the test on your own brand before the market runs it for you.
References:
- Ness Labs — Put your Meetings to Work with Chris Pedregal, founder of Granola
- Creator Economy — The 5 Hidden Rules Behind Successful AI Products
- Sequoia Training Data — Clay's Kareem Amin on Building the Sales 'System of Action' with AI
- First Round Review — Inside Clay's unconventional path to $1.25B with Varun Anand
- HubSpot Scaling Smarter Magazine — Clay's Creative Rebellion
- NBC News — Full Text of Apple CEO Tim Cook's Open Letter on FBI Court Ruling (Feb 16, 2016)
- CBS News — Anthropic CEO sticking to AI "red lines" despite clash with Pentagon
- CNN Business — Trump administration orders federal agencies to cease business with Anthropic
- CNN Business — Anthropic rejects latest Pentagon offer
- The AI Insider — Cluely Closes $15M Series A Led by Andreessen Horowitz Amid Controversy
- TechCrunch — Why Cluely's Roy Lee isn't sweating cheating detectors
- SF Standard — Inside the frat-bro startup that wants you to 'cheat on everything'