One point worth clarifying: efficient license allocation is an output of sound AI governance, not an input. The question of who gets a Copilot seat, who gets API access, and which teams justify higher consumption tiers cannot be answered by usage data alone. It requires a map of which use cases are generating value — and which are not. That map is only available if you’ve built business cases before deployment.
Most organizations assign process improvement to the people who know the process best. It seems logical. In practice, it often produces the least imaginative solutions.
Too often, I see companies invest heavily in building or buying the right tool — and then completely underinvest in making people actually use it. The result is predictable: six weeks after launch, usage has collapsed. The tool becomes a sunk cost on the P&L, and the team moves on to the next initiative. I’ve watched this cycle repeat itself too many times to call it an exception. It is the rule — and the data confirms it: according to McKinsey, 70% of software implementations fail due to poor user adoption, not poor technology.
A sales force that manages only its sell-in is not managing commercial performance.It is managing its own revenue number — and nothing else.Not the financial health of its dealer network. Not real end-market demand. Not its own 12-month sustainability. Sell-out is the truth signal. Everything else is an accounting artifact.
[Read →]
A month ago, I deployed 15 AI agents across a mid-size industrial operation, presented an AI operating model to 100 CIOs at the IDC AI & Data Summit, and ran daily executive decisions with AI as my operational co-pilot. I hired zero developers to do it. The dominant belief in 2026 — that AI transformation requires a technical team — is not just wrong. It’s the single most expensive misconception a leader can hold.
Most digital transformation projects don’t fail in production. They fail in the meeting where someone decides not to listen. I’ve seen it repeatedly across B2B industrial markets: tools designed in isolation, validated by org chart, deployed into a vacuum. The result is always the same — abandoned software, wasted budget, and a field that has learned to distrust IT. The pattern has a cost. And it’s rarely measured.