Notes on building AI that earns trust.
Enterprise AI strategy, generative and agentic AI, Physical AI and the governance that makes all of it deployable.
The body became programmable: what that does to the value of physical work.
When the body that performs the physical task becomes programmable, manual labor starts behaving more like software than like manpower.
Humanoids will arrive at work before they arrive at your home.
For embedded AI, the home is one of the hardest environments that exists, not one of the simplest.
The humanoid is not the product. The brain is.
The body may become a commodity, assembled by many manufacturers from similar parts. The brain will not.
The new illiteracy: asking the right question becomes the most valuable skill of the AI era.
AI answers a terrible question with the same fluency as a great one. It never warns you that you are masterfully solving the wrong problem.
From automation to autonomy: what changes when AI stops executing and starts deciding.
Automation asks how do I do this faster. Autonomy asks what do I stop deciding by myself.
Why emerging markets can lead the agentic AI era.
Whoever builds AI to operate in a chaotic, constrained environment builds something more robust than whoever solved it for a predictable world.
Trust is the new product: why governance will decide the winner of enterprise AI.
Capability became a commodity. The differentiator moved to a layer much harder to copy: the ability to prove the AI is safe.
The end of SaaS: why software will stop being used and start working.
SaaS charges per seat because value sits in the human operating the tool. When the agent does the work, software becomes plumbing.
AI stopped answering and started acting. Physical AI left the lab in 2026.
Physical AI is not robotics with a better chip. It is the fusion of perception, reasoning and action in a single loop.
Your home screen full of apps is a transitional artifact.
If value migrates from the app to the agent, whoever controls the agent controls the customer relationship.
Everyone wants to put AI to work. Almost nobody wants to talk about what actually blocks it.
Intelligence without governance is risk. Intelligence with governance is advantage. Trust is not promised, it is audited.
Everyone is obsessed with which AI model to use. It is the wrong question.
The model became a commodity. The real moat moved to the data, and good data is not something you buy ready-made.