What no AI has out of the box

In every enterprise AI project, the first few weeks go to the same things: integrations, permissions, security, governance. The models are ready; everything around them isn’t. Enterprise AI is waiting for its “iPhone moment” ¹: the day activating an AI capability becomes as simple as installing an app. It will come. And when it does, if every company can activate the same thing, what will set one apart from another?

Today, enterprise AI is still handcrafted. Only 37% of companies say it is already contributing to operating income, unchanged from a year ago, and barely 6% attribute at least 5% of their operating income to it ². But the market is moving toward products: 76% of enterprise AI use cases are bought off the shelf, and three providers account for 88% of enterprise model usage ³.

That path carries a risk, and it has already been measured. When Italy blocked ChatGPT in 2023, restaurants in Milan posted less on Instagram, but their posts became less similar to one another and received more likes ⁴. AI makes producing easier. It also makes sounding like everyone else easier.

Michael Porter anticipated this in 1996: the more rivals outsource activities to efficient third parties, “often the same ones,” the more generic those activities become ⁵. No company would hand its Coca-Cola formula to a standard vendor. Yet many let a standard platform decide how they communicate, how they make decisions, and how they do things.

And what makes a company different cannot be installed. A GM vice president sent his people into NUMMI, the plant it shared with Toyota, armed with cameras to photograph every square inch so they could copy it. As Toyota expert Jeffrey Liker explained, what made that plant work, the motivation and trust between the union and management, “you can’t even take a photograph of it” ⁶. The form can be copied; judgment cannot.

An invoice can be processed the same way anywhere. But every company has distinctive processes, the ones that explain why the market chooses it. That layer is strategy, not an IT matter: CEO oversight of AI governance is one of the factors most strongly associated with bottom-line impact ⁷.

When the technology matures, AI will be a product available to anyone. But one project will never stop being a project, because it cannot be outsourced: putting the company’s reason for being at the heart of its AI. Its way of communicating, deciding, and doing things: its DNA. It is every company’s Coca-Cola formula. A company that isn’t clear on it doesn’t know the formula behind its own success. One that is has the only thing no competitor will ever be able to buy.