A(I) sea of sameness.

Why every Y Combinator company in 2025 sounds the same, and how to escape the noise. Notes from a talk I gave at Future Forward, Vienna.

Back in 2023, the coffee-break question was always the same: "What are you using AI for?"

In 2025, it flipped: "What aren't you using it for anymore?"

Over the last year I noticed something else shifting alongside that, and it had nothing to do with the models. It had to do with companies. So I went down a rabbit hole. What I found became a talk I gave at Future Forward in Vienna last November.

A sea of sameness

I went and looked at the companies Y Combinator funded in 2025. All of them.

Here is one of the first I read. Klarity. Raised around 90 million dollars.

"Your Always-On AI Analyst, Advisor & Coach. Drive continuous transformation across business operations — FinOps, SalesOps, ProcureOps, anything."

Read that twice. Can you tell me what they do?

Slide showing the Klarity headline next to the question: What does this company actually do?

Another one. Unravel Carbon.

"The agentic platform for sustainability outcomes. We combine AI agents with an enterprise sustainability management platform to help organisations build enduring businesses."

Same question. Same answer.

A grid of twenty-four Y Combinator 2025 startup landing pages. They all sound the same, look the same, and lean heavily on blue and purple gradients. — A sample of Y Combinator 2025 funded companies. Source: ycombinator.com/companies

I lined up two dozen of them. They all sound the same. They all look the same. And apparently they all have a thing for blue and purple gradients. I am not making this up. Feel free to look for yourself on ycombinator.com/companies.

So the real question in 2025 actually is: how do you use AI without sounding like everyone else?

The rewrites

Here are some rewrites I did in under ten minutes. Not to boast, but to show that with some simple rules in mind, this is easily fixable.

Klarity, after:

"We spot cost spikes in your operations data in real time. So your finance team can fix them this week, not explain overruns next quarter."

Unravel Carbon, after:

"We map your supply chain's carbon footprint in a week. No guesswork, no data chasing. Most clients cut emissions by 25% within 6 months."

The difference is not talent. Just six simple things that changed between the before and after:

No subjective buzzwords. No "always-on." No "continuous transformation." No "agentic platform."
Concrete outcomes. Not "we help you optimize." Instead: "we spot cost spikes in real time."
Quantifiable results. 25% emission cuts. This week, not next quarter.
Specific pain or gain. A contrast the reader can feel.
For a specific audience. Finance teams. Supply chain managers. You know if it's you.
Simple, conversational language. If a 12-year-old would get it, you are fine.

Slide showing the six principles on the left, and the rewritten AFTER copy on the right with color-coded highlights mapping each phrase back to a principle. — How each principle lands in the rewritten copy.

The companies that get it right

At Future Forward, two kinds of people sat in the room. Founders building AI products. And people from larger companies who wanted to be seen as serious AI players, without sounding like everyone else in the AI conversation.

For both, the same question: how do you look forward-thinking and credible without buzzwording yourself into the same crowd?

Two companies show how.

ASML makes the lithography machines that produce every advanced chip in the world. Apple, Nvidia, Intel, all of their chips pass through ASML hardware. Their machines use AI everywhere: machine learning for defect detection, AI-driven process optimization, predictive maintenance. Go to their website, read their annual report. They never say "AI-powered." Not once. They are a 300-billion-euro company.

Sappi was founded in 1936. They make paper and packaging. Externally, their communication is about sustainability, premium packaging, reliability. Nothing screams AI. But look at their annual report and the picture changes: ERP systems, security infrastructure, machine learning across production and analytics, an enterprise GenAI strategy in development. AI is everywhere, they just communicate it where it matters. To investors and stakeholders who care about operational maturity. Not to customers who care about whether the packaging arrives on time and looks right.

Slide showing Sappi's AI footprint: ERP systems, security infrastructure, machine learning in production, GenAI strategy. Captioned: AI = Infrastructure, not identity.

That's the move. AI as infrastructure, not identity.

Both companies use AI extensively. Neither shouts about it externally. Their buyers don't care about the stack. They care about the outcome. If you want to be seen as forward-thinking, talk about what you deliver, not what you use to deliver it.

Three quick filters before you ship

There is no prompt that will give you specificity. But there are three filters I run any line of copy through.

The Crop Test. Screenshot your hero. Remove your logo. Remove your product name. Can someone still tell what you do? If not, your brand is doing the work your copy should be doing.

The Opposite Test. Take your headline. Replace the key words with their opposites. "We help organizations transform" becomes "We help organizations stay exactly the same." Does the opposite still sound reasonable? If not, your original is doing no work either.

The 12-Year-Old Test. Read it out loud. Would a teenager understand it? Would you use that wording when talking to a friend?

Pass all three, you are saying something. Fail any of them, you are decorating.

The line that closes the talk

AI doesn't fix vague messaging. It scales it.

If you are clear, AI amplifies that clarity. If you are vague, AI just makes you vague faster.

Get clear first. Before "scaling" anything.