The Line That Made Me Stop Reading Pitch Decks

I’ve lost count of how many pitch decks I’ve seen with “we use Artificial Intelligence” stamped on the opening slide, as if that alone were a competitive advantage. After reading the study Harvard and INSEAD just published, I can’t look at that line the same way anymore.

Because there’s a gap — structural, not cosmetic — between a company that adopted AI tools and a company that is AI-Native. And that’s not my opinion. It’s what shows up when you cross data from more than 47,000 US companies, including every Y Combinator batch from 2020 to 2024, comparing companies in the same industry with the same valuation.

What I saw in those numbers bothered me more than I expected.

The X-Ray I Wasn’t Ready For

Compared to traditional companies in the same industry with the same market value, AI-Native startups have, on average:

  • 25% fewer employees overall (in Y Combinator batches that gap reaches 30%; in mature companies from the PitchBook ecosystem, it reaches 76% fewer for the same valuation).
  • 0.5 fewer hierarchical levels — noticeably flatter organizations.
  • 15% fewer management roles — fewer managers standing in the middle.
  • 15% fewer junior roles — an almost total blackout on entry-level positions.
  • 20% more senior professionals — the team turns into a lean core of experienced people.

I sat with the practical example the study gives for a while: two companies valued at $100 million. A traditional one needs 100 people to sustain that value. The AI-Native one delivers the same valuation with 75 — and, at more advanced stages, fewer than 30.

Seventy fewer people. For the same market value. That’s not process optimization. That’s a different way of existing as a company.

What Actually Separates the Two

The part that helped me understand this gap the most was the split the researchers made between two channels of AI adoption. And I’ll admit, before reading this, I used to blur the two together in my head.

The process channel is when you hand AI tools to your team — Copilots, ChatGPT Enterprise, internal workflow automation. That makes each person more productive. But the company’s structure stays the same. Work is still done by humans, just a bit faster. It’s optimization.

The product channel is different: AI becomes the actual engine of what you sell. It doesn’t help the human do the work — it does the work, inside the product the customer buys. This is where the company’s structure actually changes, because now you can scale revenue without scaling headcount at the same rate.

I think that’s exactly the line that separates a company that just “uses AI” from one that’s AI-Native. It’s not about handing out Copilot licenses to everyone. It’s about where the intelligence lives: in the hands of whoever’s working, or inside whatever you’re selling.

Where It Hurts the Most

IndustryProduct typeHeadcount reduction in AI-Native companies
SaaS & Digital ProductsTraditional software~25% fewer employees — meaningful but moderate gain
Operational Services (legal, accounting, healthcare, HR)Labor-based processesUp to 70% fewer employees — 30 people competing head-to-head with 100

That table made me think about entire industries I know personally — law firms, accounting practices, HR teams — where growing always meant “hire more people to handle more process.” AI-Native breaks that logic outright. It delivers the outcome through an algorithm and runs on a third of the headcount of a traditional competitor. That’s not an incremental edge. It’s a different game.

What I Actually Think

I don’t read this as an apocalyptic “everyone loses their job” warning. I read it as something narrower and, to me, scarier: most companies claiming to be “ahead” on AI are actually stuck in the process channel — giving the old structure better tools. That improves efficiency. It doesn’t redesign anything.

Whoever actually captures the long-term advantage will be whoever has the nerve to rethink the product from scratch, accepting that a large chunk of execution will stop needing people in the middle. That’s an uncomfortable thing to write, because I know exactly what it means for the people sitting in that 15% of junior roles this study shows disappearing.

But pretending “using AI” and “being AI-Native” are the same thing only delays the moment your company — or your career — will have to answer that question for real.

I’m Left With This Question

What stays with me after this study isn’t “how efficient” AI makes a team. It’s “where” it lives inside the company — in the hands of whoever works there, or inside whatever gets sold. Those are two very different paths, and only one of them actually changes the structure.

Is your company today just using AI to speed up existing processes, or is it already building products where AI replaces entire steps of the operation? Let me know:

25% fewer people for the same valuation. Up to 76% fewer at advanced stages. The gap between “using AI” and “being AI-Native” isn’t rhetoric — it’s structural, and it’s already being measured.

Read Also