The Question That Changes Everything: AI-Enabled or AI-Native?
60% of companies report minimal value from AI despite significant investment. That's not a technology problem. That's a question problem.
60% of companies report minimal or no measurable value from AI despite significant investment. That's from the BCG AI Radar, January 2026.
I don't think that's a technology problem. I think it's a question problem.
Two starting questions, two different worlds
AI-enabled starts with: "How can AI help us do what we already do?"
That leads to good answers. A CRM with predictive lead scoring. A support desk with a chatbot. A reporting workflow that summarizes data instead of compiling it manually. The underlying processes stay the same — just faster, cheaper, slightly better.
AI-native starts with a different question: "What would we build today if we knew from the start that AI existed?"
That question is more uncomfortable. It challenges everything. And it often leads not to improving the old process — but to replacing it with something fundamentally different.
What this looks like in practice
A few months ago I made a decision in my own team that illustrates this difference clearly.
We needed continuous market and competitive intelligence. The classic AI-enabled answer: hire an analyst, give them AI tools so they can research faster. Cheaper than before. More efficient.
The AI-native answer was different: no analyst. Instead, an agent that runs continuously, monitors defined sources, and only alerts when something is actually relevant. No weekly reports nobody reads. Just signal, when there's signal.
That's not an efficiency gain. That's a different operating model.
The analyst would have produced reports. The agent produces decision triggers. The difference sounds semantic. In practice it's structural.
Why smaller companies have a real advantage here
Large companies have a legacy problem. Decades of accumulated processes, thousands of employees whose roles depend on those processes, ERP systems not built for AI-native workflows. AI-enabled is often the maximum achievable there — not because the technology doesn't allow more, but because the org structure blocks it.
Smaller companies don't have that problem. Fewer systems. Faster decisions. No committee that has to sign off on every deployment.
That's the Leaner Stack Paradox: fewer resources isn't the disadvantage — it's the advantage. A company building today with 50 people can start AI-native. A company with 5,000 people introducing AI almost inevitably starts AI-enabled.
The hardest exercise I know
I do this regularly. I take a core process in the company — marketing, support, reporting, recruiting, doesn't matter — and mentally delete it entirely.
Then I ask: what would we build today if this process didn't exist and AI was available as a tool?
If the answer looks almost identical to what we do today — AI-enabled is the right answer. That happens. Not every process needs to be reinvented.
If the answer is surprisingly different — then you're still AI-enabled when you could be AI-native.
Most teams I know end up in the second camp more often than they expect. Not because they lack ambition. But because the starting question has always been "how do we improve what we have" — and never "what would we build from scratch today".
That's the question that changes everything.
What's the process in your company you'd most want to rebuild AI-native — and what's stopping you?
If this was useful — more like it, every two weeks.
AI, building, and owning — biweekly, for operators who ship.