Most compliance strategies defend the wrong perimeter.

Privacy frameworks focus heavily on what companies collect.

But the commercial value of data often isn't in the collection.

It's in what the system infers from it.

For most of business history, companies built teams, processes, and headcount to execute better than their competitors.

AI is changing that equation.

When execution becomes something you can call on demand, companies stop needing to produce all the work internally.

Instead, they produce something else:

A price of zero doesn't mean there's no extraction.

It can be the most efficient form of it.

When an AI product is free, the value doesn't disappear. It can move from money into something less visible: data, behavioural signals, and dependency.

Traditional monopolies extract through price.

Most companies think they have an AI strategy. They actually have a dependency on three cloud providers controlling 63% of the infrastructure everything runs on.

This isn't a new pattern. Railroads, oil, telecommunications every major infrastructure shift created companies that started as enablers and eventually became gatekeepers.

Every enterprise security framework was built on the same assumption:

Trusted systems produce trustworthy records.

AI breaks that assumption.

Traditional security is designed to detect tampering. Checksums, audit logs, and access controls all answer the same question: "Was this record altered?"

The most influential AI law in the US may be the one that hasn't enforced a single case.

That sounds backwards.

Most people assume a regulation is effective only when companies start getting fined.

Texas took a different approach.

For 15 years, the internet rewarded one business model: own the customer relationship.

Google aggregated websites.
Amazon aggregated merchants.
Booking.com aggregated hotels.

The platform that controlled demand captured the value.

AI changes that equation.

They are the people closest to the decisions that matter. Companies don't pay a premium for code itself. They pay for three things:

• Proximity to revenue

• Control over critical systems

• Responsibility when things go wrong

Coding is the delivery mechanism. A developer who implements a specification and a developer who defines the specification may write equally good code. But they carry completely different levels of business impact.

AI is making that difference impossible to ignore. As implementation becomes cheaper, the value shifts upward toward people who can define problems, set system boundaries, make trade-offs, and take ownership of outcomes.

It s an average pretending to be insight.

Every email marketing benchmark report converges on the same output: a best time to send . But what they re really producing is a statistical flattening of incompatible contexts industries, behaviors, geographies, and intent levels collapsed into a single recommendation.

And once you look closely, the flaw becomes obvious.

A 15-minute market briefing sent to finance teams across time zones is not comparable to a promotional flash sale targeting local consumers.

For the last three years, we've been arguing about whether AI is creative.

Can it create something original?
Can it innovate?
Can it be considered an author?

Meanwhile, the legal system has been focused on a completely different question.

Who is responsible when AI-generated content causes harm?