:focal())
Use Case: Improving manufacturing processes with AI
“When can we realistically deliver this order?” is one of the most important questions in manufacturing.
The answer affects customer satisfaction, production stability, and supply chain reliability, but for many manufacturers, providing that answer with real confidence is still a challenge.
Recently, we worked with a global manufacturing organisation facing exactly this challenge. While the company had modern ERP systems and experienced planners, delivery commitments were still based on fragmented data, manual checks, and conservative buffers.
Although planners were generally able to gather the necessary information, doing so required time, cross-checking, and significant expertise.
As demand increased, so did the complexity and pressure to provide fast and reliable answers. The lack of a clear, structured, data-driven view not only slowed decision-making but also increased internal tension and led to missed business opportunities, as sales teams could not always commit to customers with full confidence.
The goal
The company wanted to move from estimated delivery dates to reliable commitments.
Their goal was clear: find the earliest possible delivery date for a new order without risking other commitments. This date needed to reflect the real situation, the actual inventory, available materials, supplier timelines, and production capacity.
At the same time, they needed to answer customers quickly. Sales teams couldn’t wait hours or days for planners to check everything manually, and conservative guesses were holding the business back. They needed a faster, reliable way to provide delivery dates with confidence, and that’s where we helped.
:focal())
The Solution
To achieve this, we built a solution using an LLM integrated directly into a Qlik Extension that combines reliable data, advanced analytics, and AI-based reasoning, within the tools planners already use every day.
Phase 1: Projecting inventory over time
The first step was to create a clear view of how inventory evolves over time.
The system starts with current stock, adds planned production, and subtracts confirmed customer orders. This makes it visible when enough finished goods will be available for a new order.
Phase 2: Validating materials and production capacity
However, available inventory alone isn’t enough. If stock is insufficient, or becomes risky later, the system automatically checks whether the required materials are available, whether suppliers can deliver missing components on time, and whether production capacity allows the order to be manufactured within the expected time frame.
Phase 3: AI as the reasoning layer
On top of this data foundation, an AI reasoning layer brings everything together. Instead of hardcoding every scenario and relying on manual calculations, the system evaluates all relevant factors (inventory, materials, supplier lead times, and capacity) and determines a feasible delivery date.
This allows planners to remain in control. They can adjust priorities, introduce subjective business rules, and respond to changing conditions. The AI does not replace their expertise, but strengthens it by providing structured, explainable support.
The Methodology
The project began by aligning and structuring key manufacturing data to ensure consistency across inventory, demand, materials, and capacity. Business rules that previously lived in people’s experience were made explicit and integrated into the system.
Only once this solid data foundation was in place was the AI reasoning layer introduced. The implementation was done step by step, ensuring transparency and user trust throughout the process.
This approach ensured that AI was not introduced as a black box, but as a practical tool that planners could rely on in their daily work.
The Result
This solution delivered immediate and measurable impact to our customer.
Faster answers: Delivery dates that previously required hours, sometimes days, of validation could now be calculated in minutes.
Lower Risk: Commitments became more reliable because they reflected real supply chain constraints.
Less Firefighting: Teams gained visibility into material shortages and bottlenecks earlier, reducing last-minute firefighting.
Better Decisions: Sales, planning, and operations started working from the same set of facts, improving alignment and internal trust.
Stronger Trust: And most importantly, planners felt more confident in the promises they made to customers.
Conclusion
This case illustrates an important lesson: AI does not fix poor processes on its own. Success depends on clear data, structured governance, and well-defined rules. When those foundations are in place, AI becomes a powerful support tool, helping manufacturers move from rough estimates to confident, reliable commitments and proactive control.
In a time where volatility is becoming the norm, the companies that succeed will not be those that simply adopt AI, but those that prepare their organisation and data to use it effectively.