Developing and maintaining a knowledge base

If you want your corporate knowledge base to provide real value to your company and serve as a solid foundation for artificial intelligence, you must not only create it but also ensure its maintenance and development.

This requires a significant organizational change in your company and a shift in your team's mindset.

What exactly do I mean? Let me explain through an anecdote.

In the companies we have audited, I was astonished to witness a recurring phenomenon that would never occur in our company.

Namely, during meetings, when new procedures were being established or important business processes were being designed, there often wasn't a person who wrote down everything that was agreed upon, turned it into an organized instruction, and saved it in a system accessible to all team members.

When I asked why such vital knowledge wasn't digitized and distributed to the team, I almost always heard the same argument:

But why? Everyone has it in their heads now and will pass it on as needed.

I shudder at such moments, thinking about all the things that could go wrong if this information stays only in people's heads. We should realize that:

  1. Human memory is fallible - it's very easy to forget significant details, especially over time.
  2. It's possible that not everyone understood the agreed conclusions in the same way - this creates a risk of misunderstandings and flawed implementations.
  3. When certain knowledge, procedure, or process changes (which happens often), it is very difficult to update it easily in everyone's heads - someone may not be present or it may be impossible to gather all the necessary people due to other obligations.
  4. There's no guarantee that a person will pass on everything 100% to others - it works like the game of "telephone," where each transmission of information to another person distorts it, and after a few such transmissions, the information becomes completely twisted.
  5. When a new person joins the team, they have to bother someone who has the knowledge - not only does this slow down the knowledge transfer process (because the new person has to wait until someone experienced has time for them), but it also significantly decreases productivity (because the experienced person is pulled away from their duties).
  6. When someone with crucial knowledge leaves the team, their knowledge is lost. It's very rare that a person leaving a company transfers all relevant information 100%, and in the case of sudden departures, the problem is even more severe and the risk even greater.
  7. In this way, we sabotage innovation and development in our company. It often happens that while reviewing corporate data and knowledge, a team member incidentally comes up with a new solution. There's no chance for such accidental discoveries if knowledge and data are not generally accessible to the entire team but remain locked in the heads of individual persons, from where they are sparingly and imperfectly doled out further.

And particularly important in the context of this guide - artificial intelligence cannot read human minds, so it will not benefit from the fact that "everyone has the knowledge in their heads."

Instead, AI excels with knowledge digitized in an organized system, and thanks to this, it can bring us great benefits.

That's why in our company, we adhere to a critically important principle:

There can be no knowledge and data in our company that are not recorded in digital form and are not available to the rest of the team.

Every team member knows that if they acquire any significant knowledge, they must immediately record it in our knowledge base. It is unacceptable for it to remain only in their head.

Maintaining the knowledge base

A knowledge base is a living organism that requires constant care. The world changes rapidly, and so does the environment both outside and inside our company. Therefore, the data and knowledge recorded in our base are constantly subject to the processes of erosion and becoming outdated.

It is necessary, therefore, to regularly review and update the knowledge base. And this cannot be done ad hoc, randomly, when someone remembers to do it. It must be structured into a process, so:

  1. It must have specific deadlines.
  2. It must have a designated responsible person.
  3. It must be verified and communicated in the form of a report.

In our company, it works as follows:

  1. Each shelf has its custodian - usually a person from the relevant team, e.g., the marketing manager is the custodian of the "Marketing" shelf.
  2. Each custodian has one day a month marked in the company calendar, when they must review their shelf and in consultation with the rest of the team ensure that all the knowledge on their shelf is up-to-date. If it is not - they must update it or delegate someone to update it.
  3. Once a month, an automatic reminder to check the knowledge base appears on the company communicator. Each custodian must then check off that they have performed their monthly check and that all the knowledge on their shelf is current.


Michal Szymanski
About author
Michal Szymanski

Co Founder at MDBootstrap , CogniVis AI and AIFor.Biz / Listed in Forbes „30 under 30" / EOer / Open-source and AI enthusiast / Dancer, nerd & bookworm.

Author of hundreds of articles on AI, programming, UI/UX design, business, marketing and productivity. In the past, an educator working with troubled youth in orphanages and correctional facilities.