Empower operations teams with AI agents that unlock smarter decisions from your ERP — no technical expertise required.
Most planning systems today are strong at structure. They understand entities, constraints, and relationships. They generate forecasts and recommendations at scale. But when it comes to interpretation, context, and behavior, the system stops.
That part is still human.
In practice, the loop looks like this:
If it feels off, they override it. The system logs the change, but learns nothing.
Over time, a clear pattern emerges: The system produces signals, and the planner does the reasoning.
For decades, planning systems have improved their models, data, and algorithms. But they have never captured the one thing that consistently drives better decisions: Experience.
The planner’s understanding of how the system actually behaves. What happens in reality.
We saw this firsthand. Even after improving forecast accuracy by around 20 percent, outcomes did not improve proportionally. Inventory was still misplaced. Stockouts still occurred. Teams were still firefighting.
The issue wasn’t the forecast.
It was the gap between the system’s recommendation and how decisions were actually made.
That gap shows up most clearly in one place: The override.
The system says one thing. The planner applies judgment.
In that moment sits the most valuable knowledge in the entire process: context, experience, and behavioral patterns.
Yet most systems capture almost none of it.
They store what changed, but not why it changed. Which means the organization repeatedly loses its own reasoning.
Knut Alicke describes this missing layer as an experiential ontology: a representation of how a system actually behaves, built from accumulated operational experience rather than static configuration.
That idea resonates deeply with what we have observed in practice. Every decision contains knowledge. Instead of treating overrides as noise, we treat them as a source of learning.
We capture:
Over time, this builds a structured representation of how the business actually behaves.
Not just its data. Not just its processes. But its lived reality.
This is what allows a system to move from storing information to understanding it.
But there is a critical constraint: This only works if it feels like a value exchange. If it feels like data capture or knowledge extraction, it fails. If it improves decisions in real time, it compounds.
As the experiential ontology grows, the system becomes better at explaining the business.
It learns behavioral patterns, surfaces context, and reduces reliance on individual memory.
But one problem remains. The future is still uncertain.
Demand shifts. Supply breaks. Policies change.
Even with perfect understanding of the past, the system still needs to act under uncertainty.
This is where the role of the experiential ontology becomes even more important. It defines the reality the system reasons over, but there is an important practical constraint.
Experiential ontology takes time to build.
It compounds through interaction, through decisions, through feedback. It becomes more valuable as it matures, but businesses cannot wait months or years to see value. They expect impact from day one.
This is where the second layer becomes critical. While the experiential ontology improves how the system understands the business over time, robust optimization ensures decisions perform under uncertainty immediately.
It does not rely on perfect knowledge. It operates on what is already available, and asks a different question:
Not “What is the best decision for this forecast?”, but “What decision remains effective across many plausible futures?”
From day one, robust optimization delivers measurable value:
As the experiential ontology matures, the system gains a better representation of reality:
And that improved understanding feeds directly into better decisions. The two layers reinforce each other.
Robust optimization provides immediate ROI, while experiential ontology compounds long-term intelligence.
One ensures the system acts well under uncertainty today, the other ensures it learns how the business actually behaves over time.
This creates a new architecture:
What emerges is not just a better planning tool, but a system that learns from decisions, reasons over context, and continuously improves how the business operates.
The next generation of enterprise systems will not be defined by better forecasts or better interfaces. They will be defined by their ability to learn from experience, because the real system has always been there.
It was just never captured.
