Envisioning a Smarter Way of Knowledge Management: Insights from a Co-pilot Demonstration

Recently, I was watching demonstration of Co-pilot, a cutting-edge AI assistant tool from Microsoft. Whilst I wasn't among the lucky members at KPMG UK to get early access, this experience caused me to reflect on how we currently manage and store knowledge within corporate environments. The specific demonstration highlighted the tool's capability to summarise meetings, draft documents, and email them to participants.

A Moment of Realisation: As I watched Co-pilot seamlessly convert a Teams call into a written document, a question struck me: Why would we still use emails and word documents as primary knowledge repositories?

This method, whilst familiar, (I get it - people used to take minutes by hand, then file them - that concept followed through even once we had computers) now seems increasingly outdated in the face of advancing AI technologies. The inefficiency of storing simple information in word documents – which are significantly larger in size compared to plain text files – becomes glaringly apparent.

1. The Wastage Behind Traditional Methods: The traditional practice of drafting minutes in a word document and then emailing them to all participants seems increasingly redundant. It's not just about the storage space – a word document with "Hello world" is 13kb, whilst a plain text file is just 11 bytes (that's 1,210 times smaller) – but also about the process. Each time an AI assistant needs to reference these documents, it has to parse through unstructured data, leading to unnecessary processing overhead and bandwidth usage.
2. A Shift to Database-Driven Knowledge Management: The alternative that came to my mind is straightforward yet revolutionary: store this content in an accessible database. Instead of sending voluminous emails that participants often ignore, why not let an AI assistant access the necessary information from a database when required? This shift can significantly reduce the data footprint and streamline access for AI tools.
3. Enhancing AI Efficiency and Accessibility: By storing data in a structured, database format, AI tools like Co-pilot could operate more efficiently. For example, when reviewing upcoming meetings, the AI wouldn't need to sift through emails or upload and parse documents. It would access a structured, concise summary directly from the database. This method not only saves time but also ensures the AI works with the most relevant and updated information, enhancing decision-making and productivity.

The demonstration of Co-pilot was an eye-opener, not just for its technological ability but for the broader implications it has on how we manage knowledge. It's time to rethink our reliance on traditional methods like emails and word documents. By adopting a more database-centric approach to knowledge storage, we can reduce wastage, enhance efficiency, and pave the way for a more intelligent and streamlined workflow in the digital age.

This was just a set of thoughts about how we can take meetings, summarise, and store the output more efficiently, but the concept can expand beyond this. Long term, do we really need contracts stored as copies in word documents when they can be assembled on the fly in the future in a more efficient manner? Most things can be broken down into just pure data and stored in a structured form. Perhaps AI will help usher in an era where this becomes the default approach to information management.