AI State of Play: What Actually Changed in the Last Quarter

The gap between what people think AI can do and what it actually does is the largest it's ever been.

It keeps coming up in every conversation I have with founders and CTOs. How fast things moved. How different the workflow looks now compared to even six months ago. So I wrote it all down.

The numbers

Last quarter's numbers tell the story:

• Spotify's best devs haven't written code since December
• 41% of all code in 2025 was AI-generated
• Cursor hit $1B ARR faster than any SaaS company in history
• Anthropic says 70–90% of their own code is AI-generated
• A quarter of the current YC batch runs on 95% AI-generated code
• Companies are hitting $10M revenue with fewer than 10 people

Garry Tan calls it the 20x startup era. He's not wrong.

What it looks like in practice

At Vandall, we have AI agents connected to our Linear board. A bug gets flagged. An agent reads the codebase, fixes it, opens a PR. We review, test, merge.

Our CTO Roger still writes code, but the majority is now AI. The role didn't disappear. It shifted. From writing code to directing outcomes and ensuring quality.

This isn't a demo. This is production workflow, running every day.

What this means for founders and CTOs

If you're leading a company, these shifts change three things:

How you hire. The 10x engineer isn't the one who writes the fastest code. It's the one who can direct AI agents effectively, review output critically, and architect systems that work.

How you think about team size. The companies pulling ahead aren't scaling headcount the way we're used to. They're scaling capability per person. Smaller teams, dramatically higher output.

How you evaluate build vs. buy. When the cost of building custom tooling drops by 5–10x, the calculus on what to build internally changes completely.

The gap most people miss

Most people are still thinking about AI as autocomplete. A better search bar. A chatbot that answers questions.

The companies pulling ahead are treating it as a team member. They're building workflows where AI agents handle the repetitive work while humans focus on judgment, quality, and direction.

That gap — between autocomplete thinking and team-member thinking — is where the competitive advantage lives right now.

Where this goes next

The last three months were the most significant shift in software engineering I've seen in my career. And the pace isn't slowing down.

If you're a founder or CTO trying to figure out what this means for your team and your product, that's exactly the kind of conversation I have every week. Happy to talk through it.