Accurate demand planning starts with reliable data, but for our client - a leading pharmaceutical manufacturer in Europe, collecting and processing customer forecasts was far from seamless. The Sales team tried to standardize forecast submissions with Excel templates, but customers used different formats or skipped the process entirely. As customer workflows evolved and non-EDI communication remained the norm, data inconsistencies and manual bottlenecks piled up. The company’s manual approach to capturing and structuring demand data became a growing challenge, impacting procurement decisions and batch scheduling.

‍

Non-standardized forecast formats

Our client needed and requested that customers submit forecasts using specific Excel templates to simplify updates in the ERP system. However, not all customers would follow these guidelines, and many submitted data in unstructured formats, including PDFs, emails, and modified Excel sheets.

‍

Evolving customer workflows

As customers updated their own ERPs and internal processes, the provided templates began to change shape, introducing inconsistencies. The lack of a unified format made it increasingly difficult to integrate demand forecasts into the manufacturer’s ERP system. Manual data entry burden over time, the variability in forecast submissions forced our client to have their employees manually input all demand data into the ERP system. This process was time-consuming, prone to errors, and led to misaligned production schedules and inefficient procurement planning.

‍

Product disruptions and inventory risks  

The inability to process customer forecasts efficiently impacted the manufacturer’s ability to plan batch production, leading to stockouts for critical medications or overproduction of low-demand items.

To overcome these challenges and provide the client with consolidated data, Nordoon deployed a Demand Forecasting AI Agent instructed based on a pre-built process template to automate the entire demand forecast flow, from data processing to ERP data updates. In this setup, the Demand Forecasting AI Agent collaborated with other AI Agents assigned to carry out specialized tasks like data extraction, formatting, quality checks, and more.

‍

AI-powered data processing across formats

We configured AI Agents to receive demand forecast data in any format, whether Excel, PDF, scan, photo or email body. The agents extracted relevant data, like product codes, quantities, delivery timelines, and customer-specific details, regardless of how the information was formatted.

‍

Data standardization with LLMs

Once extracted, our Large Language Models (LLMs) cleaned and structured the data into the format required by the client’s ERP system. This included aligning the data with internal product codes, standardizing date formats, and ensuring that all necessary fields were correctly populated.

‍

Fully automated ERP updates

The AI Agents automated the entire workflow, from receiving the demand forecast input via email to processing and updating the structured data into the ERP system through RPA rules set to ensure seamless access to real-time data. This solution eliminated the need for manual data entry, reducing errors and accelerating the forecasting process.

90% reduction in manual data entry

The automation of demand forecast data processing freed up employees to focus on strategic planning.

‍

Improved forecast accuracy & timeliness

The system stayed in sync with the latest demands, allowing employees to make faster, data-driven decisions.

‍

Enhanced production efficiency

With accurate data constantly available, batch scheduling became proactive rather than reactive.

‍

Stronger customer analysis & relations

Relieved from trivial work, employees invested more time in accurate demand forecasts, customer analysis and relations.