Why I'm definitely not building another AI platform

Perspective — Jan 5, 2026

By Piers Fawkes · 12 min read

In late 2022, AI tool-building felt limitless. Every prototype we birthed was a fresh dot on an infinite matrix—vibrant, standalone, and full of promise.

The Problem of Persistence

By 2025, the gravity of the hyperscalers—OpenAI, Google, Meta—warped the market. Specialized use cases were absorbed as fast as they were launched.

Watch as the vibrant grid turns gray and dissolves. One by one, our independent experiments are displaced into the AI graveyard.

The Value Squeeze

Specialized excellence is high effort and high margin. But "Good Enough" mass AI is expanding its reach at a rate that physically squeezes out the room.

The massive growth of the general model pushes out niche innovation. If your product is just a better UI, you're being squeezed between costs and a rising floor of standard functionality.

Vertical Absorption

The "Wrapper" is collapsing. The thin layers of UI and workflow logic are simply vanishing into the core platform layer of the hyperscalers.

The Moat Underneath

Building "on top" is a race to zero. The future belongs to those building underneath.

Large models need high-fidelity inputs. They need trusted, human-curated context. This is the Fodda layer—the strategic moat that cannot be easily absorbed.

Piers Fawkes, Founder of Fodda