From assisted planning to an autonomous operational decision-making system

Construction of a mathematical model that replicates how Novartis actually operates, enabling operational decisions to be simulated, validated, and executed with full consistency.

Novartis operates in a highly regulated pharmaceutical environment (GMP), with multiple interdependent roles, complex constraints, and a critical need for operational consistency and planning fairness. Traditionally, planning depended on human interpretation, with a heavy manual validation burden and structural risk when scaling. The Excel-based model did not allow complete scenario simulation or guarantee global consistency across the production organisation.

PLANNAM did not deploy a planner, but an operational decision-making system. A model that encodes the real logic of the organisation – constraints, priorities, sequences, and dependencies – in a structured, auditable, and reproducible architecture. This system:

  • Integrates operational memory, not just a calendar
  • Models real work sequences
  • Separates HARD constraints from SOFT behaviour
  • Generates consistent outcomes without depending on experts
  • It is not an optimiser: it is an architecture that turns rules into executable decisions.

Novartis has moved from manual planning to a model capable of simulating and validating the entire annual operation before execution. The system enables coherent, traceable, and explainable planning – critical in pharma environments – reducing dependency on experts, minimising iterations, and improving the perception of fairness across teams.

Today, the organisation has a structural foundation that:

  • Guarantees operational consistency cycle after cycle
  • Allows scenarios to be simulated before decisions are made
  • Scales to new plants, roles, and contexts

The outcome is not better planning. It is a shift in how operational decisions are made.

We do not generate plans. We structure how decisions are made.