Structured transformation for sustainable innovation
In an era where artificial intelligence (AI) is revolutionizing the business world, Infometis AG is taking an innovative approach: with an AI learning group that seamlessly combines theory and practice. Let's delve into the exciting journey of this group, which transforms AI concepts into tangible reality – all according to a well-structured playbook.
From idea to practice: How a vision became reality
It began with a simple but brilliant idea: Why not jump directly from theory to practice? Infometis AG, known for its expertise in quality and automation, founded an AI learning group with a clear goal: not only to understand AI, but to actively apply it.
An idea that counts - from employees for the whole team
What makes this initiative special is its bottom-up approach. The use cases the group is working on come directly from Infometis employees. This not only motivates them but also ensures that the AI skills learned can be applied directly in their daily work.
5-Step AI Transformation Playbook
To proceed in a structured manner, the AI learning group follows the 5-step AI Transformation Playbook. This playbook is the key to our success – and you can use it for your company too!
This is how this playbook was applied in the first use case:
Step 1: Develop and evaluate ideas and concepts
During this phase, the group gathered ideas from employees. The focus was on identifying areas where AI could offer real added value.
Step 2: Identify and describe AI use cases
Following a joint discussion and vote, a use case from the marketing department was selected, proposed by the marketing manager, Roseline. This choice underscores that AI potential exists in all areas of the company.
In this use case, the AI Roseline will help write a blog post about a video published on the Infometis YouTube channel
our Business Solutioneers gathered the requirements and recorded them in a short specification.
Step 3: Evaluate data capabilities
In this step, the team analyzed the available data for the selected use case. Specifically, the following were evaluated:
- Is there already a transcript of the video that can serve as a basis for the blog?
- How could a transcript be created if one does not already exist?
- Which AI models can directly analyze videos?
This thorough analysis helped us to identify the most suitable tools and methods for the specific needs in the next step.
Step 4: Evaluate and train AI tools
In this phase, four teams each developed a Proof of Concept (PoC), with each PoC representing a different approach to solving the challenge:
1. PoC 1 - YouTube Transcript and ChatGPT: This approach uses the automatic transcription of YouTube videos and employs ChatGPT to generate blog posts and YouTube descriptions.
2. PoC 2 - Gemini and UIPath: Here we combine Google's Gemini AI with UIPath to analyze videos and automatically create content.
3. PoC 3 - Claude and Custom Prompts: This PoC uses the AI Claude and custom prompts to generate high-quality, context-relevant content.
4. PoC 4 - Exemplary AI: An all-in-one solution that aims to generate diverse social media content from a long video.
Step 5: Implement, monitor, refine
And here we are now! Roseline is currently putting the four proofs of concept (PoCs) through their paces. Based on her feedback, the team is refining the system prompts and combining elements from the individual PoCs. It's a dynamic process of continuous learning and optimization.
Key lessons from our AI journey so far
Our experiences with the AI learning group have provided us with valuable insights that we would like to share:
1. Specialized vs. Generative: When selecting AI tools, we often face the choice between specialized solutions for specific tasks and generative tools that are more broadly applicable but require specific training. What's right for your company depends on its individual needs and should be carefully evaluated. It's worthwhile to test different options and discuss the pros and cons as a team.
2. Functional differences between the app and the API: The Google Gemini web app (Gemini – chat and be inspired) is able to analyze YouTube videos. This functionality is missing from the Google Gemini API, which we accessed via UiPath. These important differences should be considered when selecting the tools.
3. Budget planning is crucial: The budget must not be overlooked during requirements analysis. Specialized AI tools for specific use cases can range in price from cost-effective solutions to high-priced enterprise versions. Realistic budget planning from the outset helps avoid disappointment and find the right balance between functionality and cost.
4. Subjective Evaluation and AI Hallucinations: Human intelligence (HI) is essential for evaluating AI results. Especially with generative AI, results can be subjective and not always algorithmically evaluated. For example, we found that transcripts are often necessary because not every AI app or API can reliably analyze YouTube videos.
5. Teamwork makes the difference: Especially during the implementation, monitoring, and refinement phases (step 5 of our playbook), close collaboration between the business department and the AI specialist is essential. Only by combining specialist knowledge and technical expertise can the full potential of the AI solution be realized and optimal results achieved.
These insights helped us optimize our AI implementation and create real added value for our company. We hope they will help you on your AI journey too!
Conclusion: Structured innovation through learning by doing
The AI learning group at Infometis impressively demonstrates how a structured approach can make the often abstract world of AI tangible and practical. By working on real-world use cases and applying the 5-Step AI Transformation Playbook, we not only build theoretical knowledge but also gain valuable practical experience.
This initiative is a prime example of how companies can actively shape the AI transformation: with curiosity, creativity, a strong focus on practical applicability, and a clear roadmap.
Ready to start your own AI journey? Download our 5-Step AI Transformation Playbook now and begin your AI transformation – with Infometis as a strong partner by your side!
We're excited to see what other innovative solutions the Infometis AI learning group will come up with. Stay tuned – the AI revolution has only just begun!
