Organizations don't stall once on the way to AI transformation. They stall twice — at different walls, for different reasons, requiring different interventions.
That 85-point gap is not one failure. It's two walls operating simultaneously — and conflating them is why most transformation programs prescribe the wrong intervention.
Every organization deploying AI today exists in one of three states. Most believe they're further along than they are. The model names the states and the barriers between them honestly.
The mistake most organizations make is treating both walls as technology problems. Wall 1 has a technology component. Wall 2 does not. Prescribing the same solution to both is why most transformation programs fail to produce the outcome they were funded to produce.
The org purchased the technology but the culture allowed performative compliance. Pilots that never become production. Tools that get deployed but not used. Leaders who sign off on AI strategies they don't actually believe in.
The org has reached genuine augmentation but cannot cross into agentic operation because the organizational knowledge infrastructure isn't there. Agents can't execute what hasn't been articulated. Governance can't govern what hasn't been defined.
Wall 2 has a precise cause. The Knowledge Distance Problem is not a single failure — it has four distinct dimensions. An organization can be doing well on one and failing completely on another. Most are unaware of the fourth.
The IBM i community has a distinctive position relative to the Two Walls Model — one that looks like disadvantage on the surface and reveals itself as structural advantage when the model is applied correctly.
Many IBM i organizations are still working through Wall 1 — adoption that is real but incomplete, pilots that haven't crossed into production AI. That is not a permanent position. It is the cost of a culture that rightly demanded proof before commitment, and it is exactly the posture that prevents the performative compliance trap that has damaged organizations who moved faster.
The organizations that chased Wall 1 aggressively — adopting AI broadly without redesigning for it — are now discovering they built on a foundation that can't support Wall 2 work.
Wall 2 is a Knowledge Distance problem. The KD Wall requires encoded domain logic, documented decision rules, production-hardened business knowledge, and the governance structures to let agents execute against it.
IBM i organizations have been building exactly this for decades. RPG programs that encode business rules. DB2 tables that enforce referential integrity. CL procedures that document workflow. Trigger programs that execute validation logic at the data layer. The orchestration layer that Wall 2 demands — IBM i shops already built it. They just haven't connected it to the agent layer yet.
That is not a migration problem. That is the advantage most IBM i operators haven't claimed.
The Silver Lake Series applies this model in a fictional mid-market organization that walks the full transition — Coordination Tax to Layer That Doesn't Commoditize to KD Reduction Engine. The IBM i operator who reads it correctly sees their own architecture in the story.
The Two Walls Model is one section of a larger first principles framework built from thirty years of direct observation — not borrowed from other frameworks. The full document covers the nature of AI, the actual constraint, the Knowledge Distance Problem in full, the Apple Inversion, and the Human Covenant.
Read the First Principles Foundation → reggiebritt.ai