Not Artificial. Intelligence.

For three years, the world has been buying "AI." Most of it has not landed. This is not a refutation. It is a re-examination of the framing that made it inevitable.

LIVE · SCOPED EXTERNAL RESEARCH · Project NANDA · preliminary 2025 research

A category called "artificial" was always going to fail.

Project NANDA’s preliminary 2025 research reported that 95% of the GenAI initiatives it examined showed no measurable P&L impact. That result is not proof that every AI pilot fails. It is a reason to examine the framing: when intelligence is treated as a feature to procure, organisations get features; when it is treated as infrastructure, they can build a system that accumulates.

Calling machine intelligence artificial is like calling flight artificial because birds did it first.

LIVE · SEPARATE BCG POPULATION · BCG · Build for the Future 2025

The minority did not buy. They built.

BCG’s separate 2025 research classified 5% of its 1,250-company sample as future-built. That population is not the inverse of Project NANDA’s 95%. It supports a narrower point: a small group is building repeatable organisational capabilities rather than isolated tools. Perceive, model, decide under constraints, act, and learn. Build it. Run it. Let it accumulate.

The Internet was a place. Intelligence is a layer. And layers always win.