What a Fantastic Experience
Where Do I Start?
For some time now, my team and I have been discussing how to share knowledge from the various workshops, webinars, and training programs we attended over the past few years.
Of course, we share what we learn directly afterward, but we never did it as intensively as we did on April 21st, 2026.
I proposed the internal training session, and I previously wrote about my goals in my last series of posts. Those posts are quite lengthy and detailed, so I will not repeat all the specifics here. However, here are the main topics we covered:
- Linux Basics for investigative purposes
- Virtualization and secure procedures to perform OSINT
- Creating our own OSINT Tool-Repository
- Geolocation techniques
- Using AI to create helper scripts for investigations
- OSINT Team-Challenges
Their eyes were wide-open when they saw our schedule for the day
Although it sounded like a very ambitious task to fit everything into one day, we managed quite well, and the audience performed excellently.
What Did We Accomplish?
As I mentioned, it was quite bold of me to ask so much of them. So many different areas, each building upon the other. Some colleagues had semi-advanced knowledge, while others were complete beginners. One colleague told me she did not think she would handle it well because everything was completely new to her. That made the whole day a challenge for me. I could not simply drop knowledge out of nowhere without any foundational understanding to support it.
We ticked off the first two points on our agenda by distilling the necessary information down to its relevant essence for each topic. The problems we encountered on the way with Linux and Virtualization were purely technical, and we solved them relatively quickly.
Starting the creation of our own OSINT Tool-Repository including how people could make use of GitHub, along with three exemplary tools (ffmpeg, yt-dlp, and TesseractOCR) that we installed and demonstrated for their immense potential in our investigations was actually quite enjoyable.
Geolocation was something I had recently worked on, so my knowledge was up-to-date. This session served more as a reminder for the team about which tools to use and how to use them specifically.
The Fun Part: AI
Initially, I only wanted to teach my colleagues the basics of safely using AI with a CLI like Claude Code or OpenCode. In the end, we used OpenCode because it is open source and can be integrated with local models in our workflows. During the workshop, we used free-tier models available in OpenCode.
When I created the presentation, I was not sure if my colleagues would appreciate my approach of simply telling them about it and perhaps doing one or two exercises. Instead, I changed my plan. The first part was an AI Knowledge game. They were presented with 12 different AI topics of my choice, covering general knowledge, specific terms, the EU AI Act, and more, and had to share their understanding of each topic. Their knowledge in some areas was quite advanced, while in others they had never heard of the subject before. For each correct answer, they received “points of honor.” Jokes aside, they gained more in-depth knowledge about certain areas that are very important when discussing AI.
After that, we completed an AI Coding challenge with two different tasks. Participants were allowed to use agents I provided, and they had five prompts each to complete each task. The key constraint was that they could only do one prompt at a time, sequentially. This generated quite a bit of competitive spirit. I told them they would likely need only one or two well-crafted prompts to reach the goal, and in the end, one participant actually finished the second task with just one exceptionally well-thought-out prompt.
They really enjoyed the challenge, and because we had covered the basics of Linux beforehand, they very easily understood the use of the CLI and its immense potential.
Testing Their Knowledge
Of course, I could not let them leave without completing an OSINT challenge. However, because they were so fast at finishing them, we ended up doing three in total. I may have underestimated them considerably. My mistake ;-)
What Did They Learn?
Their feedback was very positive, and I believe they genuinely enjoyed all the topics we covered. My primary goal was to take away any fear of using technology they did not already know (like AI within a CLI). Instead, I gave them the toolbox and showed them how to harness it safely.
What Did I Learn?
It requires significant effort to prepare so many topics at once without overwhelming the audience while still making it a joyful experience. At the beginning, we had some technical problems I could have anticipated better. Some issues arose even though I had checked everything extensively beforehand.
However, after the workshop, when I asked the participants what they thought about the problem-solving aspect, they told me they actually enjoyed that part too. They felt less scared of making mistakes because they now know how to use AI to help solve problems in a practical way.
When Is the Next Workshop?
I hope we will repeat this knowledge-sharing session in the near future because it was truly exciting to present. I really enjoyed talking in front of my team and teaching them very important techniques for our daily work. I could easily imagine doing this more often, possibly for a different audience as well.
In the end, the law enforcement community needs to share knowledge more freely so that we can all thrive in our upcoming investigations.