The Pillar × Mechanism Grid

Definition

The Pillar × Mechanism Grid is the map that makes the orthogonality claim in Greg Kihlström‘s AI Capability Framework visible, from the book Stop Saying “AI”. It puts the four pillars — Augmentation, Insights, Orchestration, Generation — across the top, and the four mechanisms — rules, predictive, generative, agentic — down the side.

The key observation is that the grid fills in completely. Every pillar can run on any mechanism, which produces sixteen working combinations, not a diagonal. What a system does (the column) and how it’s built (the row) are two separate choices, and the full grid is the proof that they vary independently. If mechanism were just a restatement of pillar, only the diagonal would be occupied. It isn’t, which is why mechanism is an axis crossing the pillars rather than a tier stacked beneath them.

Only two of the sixteen cells share a name between a mechanism and a pillar: agentic with Orchestration, and generative with Generation. Those two coincidences are a quirk of vocabulary, and they’re exactly what fool teams into buying the heaviest mechanism whenever they hear those two words. The grid makes the trap literally visible as two cells on a board of sixteen — see the two collisions.

How It Relates to Marketing

The grid is a practical placement tool for a marketing stack. Every AI capability the organization runs occupies one cell — a pillar and a mechanism. Laying the stack onto the grid does two things at once. It shows the pillar spread (are we all Generation and no Insights?), and it shows the mechanism spread (is everything clustered on the expensive right-hand rows?).

The most useful reading is the bottom rows under the wrong columns: capabilities sitting on generative or agentic mechanisms for pillars and tasks that a lighter row would serve. Those are the forklifts, and the grid puts them where you can see them.

The Grid

Every cell is a working combination. The two starred cells are the only places a mechanism shares a pillar’s name.

Mechanism ↓ / Pillar →	Augmentation	Insights	Orchestration	Generation
Rules / if-this-then-that	Template assist	Threshold alerts	Triggered workflows	Procedural assets
Predictive	Smart suggestions	Scoring, forecasting	Predictive routing	Model-driven variants
Generative	Brand-voice drafting	Narrative summaries	Generative task planning	Authorless synthesis ★
Agentic	Research copilots	Autonomous analysis	Multi-step agents ★	Automated generation pipelines

The examples are illustrative; the point is that no cell is empty. A single pillar is reachable across all four mechanisms, and a single mechanism serves all four pillars.

How to Use the Grid

1. Place every capability. For each AI item in the stack, mark its pillar column and its mechanism row.
2. Read the columns. Check whether your pillar coverage matches your goals, or whether one pillar is crowding out the others.
3. Read the rows. Check whether capabilities cluster on the costly lower rows without the task openness to justify it.
4. Find the forklifts. Flag cells where a heavy mechanism serves a task a lighter row would clear.
5. Watch the two starred cells. Confirm that capabilities landing there did so because the task required it, not because the shared word steered the choice.

Comparison to Similar Tools

Tool	Axes	Difference
Pillar × Mechanism Grid	What AI does × how it’s built	Shows capability and technique as independent choices
Generic 2×2 matrices	Two ranked dimensions	The grid isn’t a ranking; every cell is valid
AI maturity models	Single sophistication axis	Collapse what and how into one line; the grid separates them
Feature-by-vendor maps	Product coverage	Organize by supplier, not by capability and mechanism

Unlike a typical strategy 2×2, the grid has no “good” corner. It’s a placement map, not a prioritization matrix. The value is in where your capabilities actually sit, not in migrating everything toward one cell.

Best Practices

• Use it to place, not to rank. No cell is the target. The right cell is the one the task requires.
• Expect a full board. If your stack only occupies the diagonal, you’re probably mislabeling mechanisms as pillars.
• Interrogate the starred cells hardest. Generative/Generation and agentic/Orchestration are where the vocabulary collisions land, so verify those placements were earned.
• Revisit after any re-platforming. Mechanism choices drift as vendors ship new features; re-place the stack periodically.

Future Trends

• The lower rows fill as heavy mechanisms cheapen. Expect more capabilities to drift toward generative and agentic rows, which makes deliberate placement more important.
• Tooling will auto-place stacks. As governance tools mature, they’ll increasingly tag capabilities by pillar and mechanism automatically, turning the grid into a live dashboard.

FAQs

1. What are the two axes? The pillars (what AI does) across the top and the mechanisms (how it’s built) down the side.

2. Why does the whole grid fill in? Because every pillar can be built with any mechanism. The full board proves that capability and technique are independent choices, which is why mechanism is an axis, not a tier.

3. What are the two starred cells? Generative/Generation and agentic/Orchestration — the only cells where a mechanism happens to share a pillar’s name. That coincidence drives the two collisions.

4. Is there a “best” cell? No. The grid places capabilities; it doesn’t rank them. The right cell is whichever the task requires.

5. How is it different from a maturity model? Maturity models use one sophistication axis and treat “more advanced” as better. The grid keeps what and how separate and treats every cell as valid.

Related Terms

1. The Mechanism Axis
2. The Four Pillars
3. The Two Collisions
4. The Forklift and the Coffee Cup
5. The Lightest-Mechanism Discipline
6. The AI Capability Framework (Master Stack)
7. Generative AI
8. Agentic AI

Sources

• Kihlström, Greg. Stop Saying “AI”. https://amzn.to/4wilWcA
• Greg Kihlström — official site. https://www.gregkihlstrom.com