Artificial intelligence now offers completely new approaches to rethinking knowledge management and many companies have already embraced this. Nevertheless, many companies are still unclear about how they can rethink knowledge management with AI. In this blog article, we therefore explain various approaches and present a tool as a possible solution.
Table of Contents
Challenges in knowledge management
Most of the companies we talk to can tell the following story: At some point 15-25 years ago, drives and Outlook were introduced as the first digitization activities. Over time, other systems were added: DMS systems, intranets, ERP systems and often self-developed solutions. And for some years now, more and more solutions have been offered and used as cloud solutions – the administrative burden of on-premise solutions is becoming too great. Basically, the challenges in knowledge management can be explained particularly succinctly by three points:
1. More and more data silos
Each time a new system was introduced, processes were updated and project information that was initially stored on drives was then managed via the DMS for several years. For a few years now, such documents have also been managed via Teams and SharePoint. The consequence for many companies is clear: constantly changing structures have led to a greater or lesser degree of chaos in their own IT systems. In addition, few systems are really networked with each other, but rather act as a “silo”.
2. Ever faster growing amounts of data
Due to the many digital tools, mandatory e-invoicing, working from home, generative AI and all the other trends, the amount of data in the company is constantly increasing at an ever faster rate. Exponential data growth was predicted several years ago, and generative AI is accelerating this further. Every day, even in small companies, far too much knowledge (in the form of research, documents, etc.) is created for one person to keep track of.
3. Demographic change
Another major hurdle in many companies is demographic change – until now, it was often possible to ask a more experienced colleague for help, but they are increasingly retiring and with them the wealth of knowledge they have built up. As a result, the important clue as to where which information might be located is missing. Another trend that accelerates this: Whereas in the past it was often common to stay with one employer for decades, nowadays employees change jobs more and more frequently and no longer build up in-depth knowledge at all.
As a result, companies will have to invest significantly more in their knowledge management in the future. However, this can also be an opportunity with regard to the future development of AI (see AI agents).
The basics of knowledge management
First of all, a distinction must be made between 3 types of knowledge:
- Unstructured data
Unstructured data is anything that is not stored in tabular form – i.e. documents, emails or intranet entries. DMS systems try to structure documents such as invoices in a more structured way – such data with tags is generally referred to as semi-structured data.
- Structured data
Structured data is stored in tabular form – for example in an ERP or PLM system. In such systems, characteristic values or properties of a product or a production process, for example, are usually recorded.
- Implicit expert knowledge
Last but not least, some knowledge has not yet been digitized and is therefore not accessible at all. This is usually the knowledge of employees who have been with a company for a very long time.
Solutions in knowledge management
The question that arises is how such challenges can be solved. For many companies, tidying up data volumes is not an option, as a lot of information has a very long half-life for warranty or service reasons. One of our customers calculated that he would need 5,000 person-years to structure all documents and information cleanly. The challenge: If you ask 3 people for a meaningful structure, you often get 3 different answers – so there is no meaningful sorting measure for an AI that would allow it to automatically structure information.
In order to leverage the knowledge contained in structured data (e.g. to answer questions such as: “How much sales did I make with customer X in period Y with product Z?” or “How many resources did I need for process step X?”), business intelligence solutions are the means of choice, as they can analyze and extract this data very well with algorithmic capabilities.
Implicit expert knowledge or analogue knowledge initially faces the challenge of having to be digitized. To do this, we recommend the following process as described in this blog post:
- preparing expert interviews before employees leave on the relevant topics
- recording audio recordings, e.g. in the form of a conversation
- transcription with the help of AI
- writing knowledge articles based on the transcripts with the help of AI
The result is primarily unstructured information in the form of blog articles or knowledge documents.
Knowledge management with AI and unstructured data
This leaves the largest “block” of information that has not been solved well for years. A few years ago, intranets were still being used to create “knowledge hubs” or similar, but it has now been realized that intranets are another data silo that needs to be maintained and therefore contributes to the problem.
Thanks to AI, companies now have the opportunity to make internal knowledge quickly and easily accessible with the help of intelligent search solutions. One approach is the so-called enterprise search approach: instead of storing knowledge anew, it is left exactly where it is. Using so-called connectors, an AI-based search is used to search the internal data pools – similar to “Google”.
Modern enterprise search solutions can use large language models (LLMs) to search through large amounts of data much more efficiently and quickly than was previously possible. An example of a possible enterprise search solution is shown here:
Of course, such solutions quickly raise questions about IT security – e.g. how are existing access rights handled, where is data hosted, etc.? These are all questions that we are happy to answer for our customers.
How to get started with knowledge management with AI?
amberSearch provides a free demo in which over 10 different systems are connected. The demo can be tried out free of charge here:
How to get started with AI-based knowledge management?
If you want to get started with knowledge management, you should first try to get an overview of the challenges facing your own workforce. In this blog article, we have provided an example of what a survey could look like as an initial approach for a company. You should also think about the business case as soon as possible – we have also written a blog article on this and there is also an ROI calculator on our pricing page.
We would be happy to support you with the use of AI – send us a contact request here: