When AI shows up as a pillar in a strategy diagram, ask whether the rest of the diagram could function without it. If not, you do not have a pillar. You have a substrate, and substrates do not go in the pillar row.
It was a Friday afternoon and my cross-functional partners and I were looking at a strategy doc together. A real one. The kind that a Business Unit prepares when it wants to show the C-Suite and the board that it has thought hard about the next three years. The document had pillars. Five of them, arranged in a tidy horizontal row, each with a confident two-word label and a paragraph of supporting prose underneath. One of those pillars was “Agentic Automation,” which is the new way of saying “AI that does stuff,” and which was being treated, in the document, as roughly equivalent in importance to “Customer Acquisition” or “Revenue Operations.”
As we reviewed it, we talked about how something felt off, but couldn’t name it for a second, the way you sometimes can’t name what’s wrong with a sentence until you read it out loud and hear the missing comma. Then my Engineering partner had it. Calling agentic automation a strategic pillar is approximately as useful as calling “having a database” a strategic pillar. It feels true at the surface. It is also a category error.
This is the test I am offering you, and I am calling it the Database Test, because I am not above naming things after the first metaphor that comes to a leader late on a Friday. The test goes like this: when you put a thing on your strategy diagram, ask whether the rest of the diagram could function without it. If a pillar can be removed and the building stands, congratulations, you have a pillar. If a pillar can be removed and the building reveals itself to have been a hologram all along, you do not have a pillar. You have a substrate. Substrates do not go in the pillar row. They go underneath all the pillars, holding them up.
Most product strategy decks I have seen in the last eighteen months make the same mistake. AI shows up as a pillar. Mobile, ten years ago, did the same thing in a thousand decks. Cloud did it before that. Search did it before that. The pattern is consistent: a foundational shift gets mistaken for a strategic option, because at the moment of the shift it is the only thing anyone wants to talk about, and the natural place to put the thing everyone wants to talk about is in the place where a strategic option would go. So the foundational shift sits there, posing as a pillar, and the diagram gets less true every quarter as the rest of the company gradually realizes that the new substrate is, in fact, in the floor, the walls, the wiring, and the air conditioning.
The fix on the document I was looking at was small and physical. We removed the pillar, and we put a horizontal bar across the bottom, running underneath every other pillar and labeled with the substrate in question. It was a five-minute change. It also reorganized everything above it. A pillar promises a roadmap. A bar promises a posture. A pillar implies an investment. A bar implies an obligation. A pillar can be sequenced and resourced. A bar has to be everywhere all at once, which is uncomfortable for organizations that want to phase things, but which happens to be true.
If you are a design leader, the practical move is this. Get hold of your company’s strategy document, the most recent and most important one, the one that is actually being used to allocate budget. Find every pillar that contains a word like “AI,” “data,” “platform,” “automation,” or “intelligence.” Run the Database Test on each. For every pillar that fails, redraw it as a bar. Then look at what is left in the pillar row, and ask whether you can finally see the actual strategy now that the substrates have moved out of the way.