Expertise has been growing in companies for years. However, in SMEs in particular, much of it is in the heads of employees – or in hundreds of folders in the paper archive in the basement. Before you can deal with “AI”, the knowledge must first be digitised. But what is the best way to do this?
First of all, we keep realising that far more knowledge is already digitised than we think. Just because it may not have been structured or stored editorially does not mean that the knowledge is not already digitised and usable “between the lines”.
Furthermore, there are different types of digitised knowledge: knowledge that is structured (e.g. in SQL tables), completely unstructured (any documents, folders, messages, …) or semi-structured (e.g. editorially maintained intranets).
Why digitise analogue knowledge?
There are many reasons to digitise knowledge. Some of the reasons are as follows:
- Demographic change: especially in the younger generation of employees, everyone is used to having information directly via their mobile phones. They will therefore not want to go to the archive or ask five colleagues for information. If you don’t digitise knowledge and information or have good processes in place, you won’t be seen as an attractive employer.
- Shortage of skilled labour: The baby boomer generation is retiring and taking their knowledge with them. This means that companies are losing valuable expertise and it remains out of reach for young/new employees.
- Change in the labour force: In the past, it was particularly important to come out of training with knowledge. Nowadays, it is much more about transfer performance. A person has to make decisions based on the right information and needs as comprehensive an information base as possible to do so. This requires different tools, but also a rethink in the way the company works.
- Increasing efficiency: Scaling effects can only occur with digital tools. The more efficient I want to become, the more digitised and automated I need to be. I can only achieve this if the necessary expertise is available, because in future it will be about doing more with fewer staff.
The knowledge from the minds of the employees is therefore one of the most important pieces of knowledge a company has, because it is the knowledge that makes a hidden champion a hidden champion. This includes, for example: How are machines set up, How do the processes work, What experience was gained from which projects, etc…
When we talk to established SMEs, we often hear them say: “We don’t need to talk about AI yet, we haven’t digitised enough know-how” – and that’s in 2024. That’s why we want to give all these companies a best practice in this article on how analogue knowledge can be digitised as easily as possible. Ideally, such a project should be part of a digital strategy.
Digitising knowledge from the minds of employees
Most digital tools build on existing digital tools or process digital information better. In order to do this, however, the knowledge must first be digitised. For most machines, it is easier to deal with structured data, but in this case it is primarily about unstructured knowledge – which is also not so easy to digitise. After all, very few employees simply sit down and write down everything they know. We would therefore like to recommend the following process, which can also be easily “tried out” using various tools.
Firstly, of course, you should define what you want to talk about and which topics you want to address. It may well be that you have to “lock yourself away” with an employee for 2-3 days in order to capture all the expertise from X years of professional experience in interviews.
- Record a podcast/interview
- Objective: Employees share their challenges, tasks and experiences.
- Purpose: Very few people will be able to simply write down their expertise in a reasonable amount of time – because they don’t like writing, because it takes too long, because they simply don’t like working on the computer, … . In a (moderated) interview, however, these people can talk about exactly the topics that need to be digitised. Where necessary, existing experience or process knowledge can be explored in greater depth.
- Measures: Ensure an appropriate technical setup (high-quality microphones, software for recording separate audio tracks).
- Tips: Record the audio tracks of the interview partners separately so that you can transcribe them as individual audio tracks if necessary. Make sure that an appropriate level of detail is recorded.
- Transcribing the digital audio track
- Objective: Convert the audio track into a digital text
- Purpose: Nobody wants to spend ages listening to long interviews to access the know-how. With the help of a transcription tool, you can make the information readable.
- Measures: Use high-quality tools such as Whisper & Co (there are various providers) that understand German. Help from IT/external parties may be necessary here to get the transcription in the appropriate quality
- Tip: There are already good open source models here, some of which are able to identify different speakers or names so that hardly any reworking is required. With open source models, only the server capacity has to be paid for.
- Generation of knowledge articles
- Objective: To turn an interview into a continuous text or knowledge article
- Purpose: A clearly structured knowledge article is much easier to read than an interview, which may jump between topics or ask questions.
- Measures: Use tools like ChatGPT (better the paid version) or similar tools that use generative AI to summarise the information. However, ChatGPT has the challenge that it is not GDPR-compliant in every form. With amberAI, we offer an alternative that companies can use to manage internal company expertise with AI in a GDPR-compliant manner – across various internal systems.
- Tip: Try out the so-called “prompt” beforehand. The more specifically you tell the AI from which perspective or what it should write about in which tone of voice, the better your result will be.
- Correction
- Objective: No AI is perfect – so you will have to correct it manually.
- Purpose: Check the generated texts for content accuracy or add to them where necessary so that they are complete. Ideally, the experts who want to/should share their expertise with the team should also do this.
- Archiving of digital knowledge
- Objective: Knowledge must be accessible
- Purpose: The entire process only makes sense if the knowledge is subsequently made available to all employees. It should be ensured that the knowledge is stored in such a way that employees can easily access it.
- Measures: Store information in dedicated tools – intranets, DMS or other file storage systems.
- Tip: If you end up with dozens, hundreds or even thousands of such articles, nobody will know exactly where everything is afterwards. That’s why there are tools like amberSearch that use AI to help you access exactly the right information.
Digitising analogue knowledge from paper files
This blog article primarily dealt with the digitisation of knowledge from employees’ heads. However, a large proportion is of course still digitised in archives, which can often be stored (and made usable) much better digitally.
There are various scanning service providers who digitise the archive and also use optical character recognition (OCR) to make the digitised documents computer-readable. And building on this, technologies such as generative AI can of course also play to their strengths for SMEs. If you would like to gain some initial experience of what is possible, you are welcome to try out our online demo. We have connected over 10 different digital data silos in our amberSearch solution and let our AI loose on them:
Conclusion
There is no tool for digitising knowledge from people’s heads (for the foreseeable future), but there are processes and best practices for digitising knowledge as easily as possible with the help of AI. If you are interested in using AI in your company, you are welcome to read through our other blog articles.