Knowledge management is a burnt term in many companies. However, methods in the field of AI are opening up new opportunities for companies that are worth using ai in knowledge management. In this blog article, we describe what went wrong in the past, why it’s time to take another look at knowledge management and what has changed as a result of AI.

Why is knowledge management a burnt term?

Especially in the 2000s, knowledge management was a hyped term. The ability to store information digitally and find it again quickly suddenly became accessible to many companies.

In practice, new data silos were created in companies – and still are today. Drives, SharePoint, Teams, intranet, DMS, etc. – to name just a few. Even back then, it quickly became unclear where which information was stored.

The solution back then was to simply use a knowledge management tool and store all relevant information there. Then there is always a central place to find the information.

The problem with this is that such a tool always needs to be maintained. Far too often, such projects have been about the processes, the technology or the tool itself. The employees or the necessary change in employees was simply missing and many still stored things in the data silos – just as they did in the past.

Last but not least, it should also be noted that a large part of the knowledge is not in the strictly documented processes, but in the daily doing. In the documents, emails, etc.

If you would like some tips from us on how to digitise analogue expert knowledge from the minds of employees, you should take a look at this blog post.

Why does knowledge management through AI deserve a new chance?

A lot has changed in recent years. In addition to the culture, the technology has also changed. So here are 5 reasons why it’s worth taking a fresh look at knowledge management:

1.       Everything is more digitalised

Where companies in the 2000s stood on a green field and had relatively no data, most companies can now look back on 20+ years of data. This means that significantly more expertise has already been digitised. This therefore does not have to be digitised for such a project.

2.      Interfaces are state of the art

While interfaces were still foreign words for many software providers in the early 2000s, almost all systems now have interfaces. This has significantly simplified communication – and also the possibility of extracting knowledge from the systems. Solutions that do not have interfaces are now the exception rather than the rule and will probably disappear from the market sooner or later.

3.      The culture in companies is changing

Whereas some employees used to have the self-image of sitting on the know-how and the data of the departments was strictly separated, this has loosened up in recent years. Collaboration tools such as intranets or teams, for example, have made information much more accessible and companies are also focussing on collaborative work. Employees are also realising that it is becoming increasingly important to share knowledge and make it accessible.

4.      Data volumes are increasing exponentially

If you ask your IT department how quickly your company’s data volumes have grown over the years, you will probably get answers of around 50% or more. So not only is the technology there, but it is also necessary to introduce AI tools in order to maintain an overview of the expertise.

5.      Employee throughput & skills shortages have increased

Whereas in the past it was often the norm for employees to stay with a company for several decades, sometimes from training to retirement, younger employees in particular change jobs every five years at the latest. In the past, expertise stayed with the company, but now it leaves the company all the more quickly.

Now more than ever, companies need to ensure that new employees become efficient as quickly as possible and utilise the company’s expertise correctly.

How can AI be used in knowledge management?

Since 2022, generative artificial intelligence in particular has been experiencing real hype thanks to applications such as ChatGPT.

However, the technology behind it has been around since 2017 and has several use cases that can be ideally combined for knowledge management and offer many other possibilities.

The use cases of information retrieval and generative AI are particularly relevant for knowledge management.

What is information retrieval?

Information retrieval is about finding information from a specific data set. Of course, this can also be done without artificial intelligence and is then called keyword-based search. But a lot has changed: in 2017, Google presented so-called transformer models at a developer conference, we also open sourced some of our models in 2020.

This means that you no longer have to search for word frequencies in a document, but can now search for information using semantic content.For employees, this means that it is no longer important to find exactly the right wording, but above all the meaning of their search query. In this blog post, we have written down how an AI-based search works in contrast to a keyword-based search. In a nutshell, these are the biggest advantages of this technology for information retrieval:

  • Information or data no longer needs to be edited or manually prepared to be understandable for the AI.
  • The AI understands what the documents are about. It is less about the exact wording and much more about real content.

Ideally, this technology should therefore be used in an enterprise search. An enterprise search is a company-internal, cross-system search engine that helps employees to find internal company information.This means that “operational” information from documents, team chats, project management tools etc. is also included in the search without having to be processed separately.With amberSearch, by the way, this all works in compliance with access rights and the GDPR.

Generative AI in knowledge management

Generative AI alone is not much use in knowledge management; it must be combined with enterprise search.Generative AI can then better prepare or summarise the information that is relevant for the employee in the respective situation. Such a system is then called a Retrieval Augmented Generation (RAG) system.

In this case, the input of the generative AI is defined by an upstream search. All access rights can then also be taken into account. The combination of enterprise search and generative AI gives companies the opportunity to build a GDPR-compliant chatbot, similar to Microsoft’s Copilot, but across all internal company software systems. If you want to get a first impression, you should try out amberAI.

AI will be indispensable in knowledge management

The fact that AI is now so advanced that information no longer needs to be processed means that the introduction of such systems is no longer a huge project. For example, the technical introduction of amberSearch as a managed service solution ideally takes 2-3 hours by an IT administrator.

At the same time, data volumes are increasing so rapidly that even today no employee has an overview of all the information and the shortage of skilled labour is intensifying, so companies will have to introduce solutions that make their employees significantly more efficient.

This leaves the employee time to deal with the customer’s problems and the further development of processes instead of having to deal with the procurement of information within the company.