100% AI-Written Code

The Avengers. That's what we call ourselves; we're sort of like a team.

Over the last 6 months, product development within enterprise software changed. We transitioned from 0% of new code written by AI to nearly 100%. And more recently, every company’s board and management team we work with is already there.

Once again, the biggest bottleneck is human-in-the-loop PR reviews and manual QA, but that will get figured out shortly, at a planetary scale (see Chasing Bottlenecks).

Roadmaps have been accelerated by quarters, and in some cases, years. Agentic workflows in the product and agentic development in the SDLC is now common across the landscape.

Some companies are further ahead than others, with existing AI-native, end-to-end platforms built from the ground up. Others are still struggling with legacy acquisition-integrations and roadmap kerfuffles.

But one thing that isn’t yet common, but we’re likely to see as the next major industry-wide hill climb, is the transition to developing and owning your own proprietary models in-house. After all, proprietary data is only worth ~10% of the commercial value of customer-facing products. And driving defensible growth means you have to use that proprietary data for something novel that your competitors don’t have.

Of course, the chess match continues when competitors get similar types of data and begin creating similar types of models.

Once again, the question is how do you maintain persistent difference when everyone fast-follows each other with only a day’s lag?

That’s what we’re good at.

—Sean