Cronos AI Lab, where Ideas turn into Projects

The tech industry has spent the last few years talking about the massive potential of artificial intelligence. Now, the conversation is shifting toward actual execution. Companies are moving past the experimental phase and looking for practical ways to integrate these tools into their daily operations. A recent showcase highlighted this shift, demonstrating how organizations are moving away from slow, manual processes and adopting smart, AI-assisted workflows. The goal is simple: automate repetitive tasks so human workers can focus on higher-value decisions.

‍

The Lab's Mission = Unlocking AI's Potential

The AI Lab is designed to operate as a space where complex business hurdles are turned into functional software. That is the core mission of the project. Yorrick Schoonheydt, who opened the session, pointed out that this isn't a physical room packed with robots. Instead, it operates as a specialized research and development hub. It serves as a destination where teams from Cronos can bring their specific challenges and find AI-driven solutions without getting bogged down in endless technical research themselves. While the lab currently focuses on internal projects, the long-term plan is to offer this setup to external clients. This will allow businesses to test and build custom AI tools tailored to their needs, ensuring good ideas actually turn into working applications.

‍

Smart Solutions in Action: Automating Sales and Car Inspections

The showcase highlighted two clear examples of this technology in action. First, Alban Capaj introduced Leadflow, an AI-powered sales outreach platform. Alban noted that traditional prospecting, which involves finding target companies, identifying contacts, and drafting custom emails, takes up a massive amount of time. Leadflow automates these steps. The tool helps users identify target organizations, pick the right contacts, and draft personalized email campaigns using generative AI. Crucially, it keeps a human in the loop to review and approve the final output before anything gets sent. The entire system runs on a lightweight, efficient backend powered by the low-code N8N framework.

Next, interns Sepp Verbuyst and Mitko Todorov presented the Cronos AI Damage Inspector. This tool aims to automate the tedious process of inspecting leased vehicles, a task that historically required painstaking manual checks by specialists. Sepp pointed out that the team had to overcome several real-world hurdles during development, including inconsistent image quality, bad lighting, and glare on car surfaces. The AI system processes vehicle photos and flags damage. Mitko explained the underlying technology: the system takes high-resolution images, breaks them down into smaller tiles, and uses an object detection model to find physical damage. From there, a visual large language model confirms the issues and estimates the size of the damage, speeding up the entire inspection process.

‍

The Road Ahead

Both projects are still evolving, with clear upgrade paths ahead. For the damage inspector, Sepp and Mitko outlined several updates designed to make the tool more robust. The team plans to migrate to Azure Blob Storage to handle images more efficiently, as the current Spark-based storage setup consumes too much space. They are also working on an ensemble voting system, which uses multiple AI models to analyze damage collectively, raising overall accuracy.

Another interesting update is the introduction of a rental norm ruler. This physical reference tool is placed next to a scratch or dent during a scan, helping the AI accurately calculate the exact size of the damage, which has been a tricky technical challenge. The system also includes a feedback loop. This lets human inspectors correct false positives or flag missed damage, which helps retrain and sharpen the model over time. This feedback loop ensures the software continues to learn and adapt to new real-world data.

‍

Conclusion

The latest projects out of the AI Lab show exactly what practical AI looks like today. Whether it is streamlining sales outreach with automated campaigns or using computer vision to assess vehicle damage, these tools prove that AI is at its best when it supports human decision-making. By combining low-code frameworks with advanced machine learning models, the lab is proving that the path from a smart idea to a working business tool is shorter than ever.

‍

Curious how AI can transform your operations? Contact us to explore tailored solutions!

‍