01Business problem
Which specifications fit a market, how a suffix change affects CO2 impact, how a price or positioning shift affects the product mix: every answer meant navigating multiple tools and holding planning logic as tribal knowledge instead of standardized work, since no tool could apply planner judgment, only surface data. Decisions were slow and inconsistent across planning teams.
02Intervention
We started at the gemba, listening to planners explain their logic, constraints and pain points, and translated that into a first proof of concept. Value was validated in testing and feedback sessions, and the tool improved in cycles of iterative development and improvement until it reached its minimum viable accuracy.
03Role of data & AI
An AI agent trained to reason like a planner, not query like a database: it generates planning insights that respect the same business rules, suffix logic and market constraints a planner would apply, and answers natural-language questions with recommendations, explanations and scenario insights grounded in the enterprise planning knowledge.
04Adoption
Over time the planners took ownership of the knowledge: a continuous improvement loop where they define and maintain the business rules and business glossary, while we own the backend and scale the tool up step by step.
05Measurable impact
Around 35% higher planner productivity and efficiency, roughly 90% time reduction on manual insight generation and faster, more consistent decisions across more than 400 planners.