At a high level, knowledge management consists of two parts - maintaining and distributing data. The sophistication of Large Language Models (LLMs) has presented a clear opportunity to enhance the distribution side of knowledge management, and the recent AI craze has software companies scrambling to capitalize on that opportunity.

An AI technique called Retrieval-Augmented Generation (RAG) fits the challenge of knowledge distribution like a glove. If an employee at your organization has a question, you can send the question to ChatGPT with some relevant data from your knowledge base, and you will likely receive a good answer. Dozens of companies are using this method to solve knowledge distribution. One-half of knowledge management is in a good place.

But what happens when you share outdated or incorrect information with the AI? What happens when you don’t have any relevant data to share? You get a confidently worded wrong answer, or, in the best case, no answer at all. These are problems of knowledge maintenance and generation, which are not nearly as straightforward to solve with a quick message to an LLM. The solutions to these problems lie within the people of an organization.

At PairUp, we are building a platform that addresses both sides of knowledge management. Knowledge management systems cannot provide full value unless the question-and-answer loop incorporates rich information from people’s minds. Our approach to Human-Augmented Generation focuses on empowering people to contribute more with less effort. Doesn’t that sound better than talking to a robot full of stale data every day?

Andy Garvin

PairUP Co-Founder & Chief Technology Officer