Knowledge of systems, especially legacy systems is held with experts who have been with the company a long time. When they retire are you sure that the time given to them to transfer that knowledge has achieved the goal?
When experts retire and responsibilities pass to remaining employees or new hires, the burden of older systems can become immediately apparent. With new tools to help workers with their goals in the form of AI assistants and with a shift in our view of how knowledge can be organised, it might be the right time to look at the process and culture around this critical aspect of business.
Your more mature experts have most likely earned their chops over an extended period of time, sometime decades. It can be difficult for them to condense all that learning into a document prepared a few months prior to departure, given all the other demands on their time.
By structuring a specific role for these ‘knowledge stewards’ and shifting their responsibilities you gain a couple of advantages:
With less time in their old role the new employee must step forward and start to take the reigns, providing an opportunity to onboard to the role whilst the expert is still with the company
Your expert can begin to build documentation and learning tools in a structured way that will be available to the organisation. This structure, as we will see, must align with data quality policies for Artificial Intelligence. Although, it also allows flexibility for the expert to furnish documentation with unstructured data as long as the data catalogue is kept up to date
Organisations that are embracing data quality for analytics and insights will already be aware of the requirements on their data. Data must be well organised, catalogued, clean and reliable. Building a single source of truth and democratising access to that data provides cohesion and the trust everyone needs to be able to rely on what the data is telling them.
Data of this kind often lives in a data warehouse/lakehouse and is stored with all the meta data to provide machines with the context they need to understand it. It is not too much of a leap to extend this strategy to knowledge stores so that AI models may be trained on it or LLMs tuned from it so that the knowledge may be extended to the rest of the organisation.
By implementing tools for employees allowing them to access this knowledge via prompt interfaces whilst executing tasks, onboarding and problem solving can be optimised greatly.
Off course there are implementation details that I have not covered, but the premise of this article is to begin to think of expertise not as a retiring resource, but as a different class of employee that can become a knowledge worker, working with the new advanced tools to build the data and structure for the next generation.
