Summary
Liam Ottley, who runs an AI transformation company, argues that most businesses should pause spending on AI systems for about a year, even though he sells exactly that kind of service. His reasoning is that nobody, including the labs, startups, and his own agency, has yet worked out what an AI-native company looks like, so many custom builds are improvised and fragile.
He uses the early car industry as an analogy. Henry Ford had a working engine in 1896 but needed years before the Model T made cars practical. Today, Ottley says, the “harness,” meaning one strong model with a set of tools, as in Claude Code or Codex, has won out over training custom models or networks of small ones. But the harness is only the engine. He calls the current phase “the waiting room” before an all-in-one AI platform for business arrives.
Ottley breaks that platform into seven pieces, including a context layer of organizational and role-based documents, a data layer or “second brain,” and agents acting as AI employees. He says no company has shipped all seven. He also warns that these platforms are all-or-nothing by design and carry heavy switching costs, so the next one to two years will be a land grab with discounted onboarding. His advice is to go all in when you choose a platform, while agencies should specialize in installing one deeply.
📺 Source: Liam Ottley · Published October 19, 2026
🏷️ Format: Opinion Editorial
