The Challenge

A global pharmaceutical organization manages a vast library of marketing materials, ranging from brochures and presentations to digital assets. To support efficient reuse and compliance, these materials must be accurately classified based on content, therapeutic area, and intended messaging.

Previously, this classification was done manually, a time-consuming and error-prone process that often resulted in incomplete or unclassified content, limiting the ability to search, retrieve, or repurpose materials across teams.

The Solution

Common Sense AI designed and deployed a scalable automation pipeline that uses large language model (LLM) technology to classify marketing materials based on their contents and context.

Key features of the solution included:

• Batch processing of more than 1,000 materials at a time

• A deterministic LLM output format to ensure consistency and ease of integration

• An intelligent recovery mechanism: when the model returns an incorrectly formatted response, the system engages in a follow-up conversation with the LLM to resolve the issue automatically

The pipeline was engineered for robustness and scalability, enabling fast and reliable classification even at high volumes.

The Result

With the new classification system in place, teams now benefit from:

• Automated tagging of materials with accurate categories

• Reduced manual effort, freeing up time for higher-value work

• A scalable solution that adapts to any new or incoming batch of assets

• Improved findability and reusability of marketing content across the organization

This project highlights how LLM-based automation can streamline knowledge organization in regulated industries, improving both speed and consistency at scale.