AI can be a powerful ally in the workplace, helping employees find the information they need within seconds instead of digging through stacks of paper or endless digital documents. But for industries like pharmaceuticals, where operations are tightly regulated and data is highly confidential, using popular open AI tools is simply too risky – and often strictly against company policy.
This was exactly the challenge faced by our client, GE Healthcare’s factory in Norway. They needed faster, easier access to critical knowledge without compromising security. Our solution? A local AI-powered knowledge base powered entirely by internal data, operating on a local server. No one outside the company – not even us – had access to the documentation. The result: streamlined daily workflows and quick, reliable answers – all while keeping sensitive information safe.
Background: When Critical Know-How Is Hard to Find
In their daily operations, industrial companies are overwhelmed with vast amounts of documents, reports, and correspondence that have often been collected over many decades. Moreover, these documents are frequently misnamed, poorly organized, or incorrectly categorized.
Finding the right information often requires painstaking searches across multiple sources, contacting colleagues and suppliers, and manually piecing together facts.
Imagine a maintenance manager who needs to quickly check: “Which parts were replaced during the last inspection of machine X, who performed it, and what were the technicians’ recommendations?” Without the right tools, finding the answer can take hours – reviewing PDF reports, locating related Excel files, checking emails, and asking colleagues for audit notes.
The Solution: Finding the Right Information Quickly and Easily
For a GE Healthcare factory in Norway we created a system being a solution to this and similar problems. It is an intelligent knowledge base based on artificial intelligence and large language models (LLM). By combining knowledge from multiple internal sources, like extensive pharmaceutical documentation, it delivers answers in just a few seconds, through a convenient web interface in the form of a conversational chat, while showing where the data comes from.
Users can ask questions in natural language – English and Norwegian, and the chat provides responses based on verified sources, clearly indicating the documents used to generate each answer. This allows employees to make faster decisions without frustration or wasted time, helping the organization operate more efficiently and effectively.
