We understand the importance of integrating fresh talent into our teams. Students bring innovative perspectives and the latest academic insights, which we value greatly. Recognising the need for in-depth research and practical execution, our product team sought an ambitious student to contribute meaningfully to our projects.
With AI increasingly integrating into both professional and personal workflows, we were eager to explore its potential within Fyndoo—understanding not just what was possible, but also where its limitations lay. Given that a typical graduation assignment spans around four months, we needed a well-defined scope that allowed room for exploration while ensuring the student had the time and space to experiment and develop their skills.
This led us to search for someone who could dive into the potential of Retrieval Augmented Generation (RAG) architecture.
“When drafting a project, we always focus on three primary goals: the technical aspect, identifying talent, and exploring exciting new possibilities with our product.”
Welcoming Cindy
In September, we had the pleasure of welcoming Cindy Ye to our Fyndoo team. Currently pursuing a degree in Software Engineering at Windesheim University of Applied Sciences, Cindy was eager to take on a hands-on project where she could put her software implementation skills to the test. Her passion for AI and its transformative impact on business made her a perfect fit for our team.
“When choosing a company, I was looking for a challenging project, but also a place where the mission and values aligned with mine.” — Cindy Ye
We believe a successful internship or graduation project starts with a strong foundation. Our onboarding process happens in two phases:
- Getting to know each other—creating a space for both the company and the student to align on goals, expectations, and working styles.
- Technical onboarding—immersing into the project’s scope and objectives to ensure a smooth start.
We understand that every student brings a unique skill set and perspective, so we tailor our graduation projects accordingly. With clear guiding principles in place, Cindy’s challenge was set: she would focus on building an AI-powered assistant in the form of a chatbot to streamline the configuration process for our Lending platform (BLP). This innovative solution aims to simplify user interactions while also exploring how AI can enhance internal processes at Fyndoo.
The model
The user interacts with the AI Assistant via a chatbot (text-based). The message is then sent to the AI Assistant, which processes it and generates a response. But how does the AI Assistant actually understand our Lending system? The truth is—it doesn’t, at least not initially.
The AI Assistant is powered by a Large Language Model (LLM), pretrained by Microsoft Azure’s Open AI. To make it more relevant to our Lending platform, we retrieve and attach relevant data before processing the request. This approach—known as Retrieval-Augmented Generation (RAG)—helps generate more precise responses.
One key challenge was that we didn’t have direct access to structured data within the database. Since this information wasn’t stored, we needed a knowledge base. This database was built using existing manuals, which were processed through a data ingestion pipeline. The pipeline vectorizes the content, making it easier to retrieve relevant information based on user queries.
This is just the beginning—there are many ways to enhance the AI Assistant further. While the current implementation focuses on providing helpful information, in the future, we aim to expand its functionality so that it can assist with direct configuration tasks within the system.
“I have already some experience mentoring students, but what makes Cindy so special is her capacity to listen, learn and implement new knowledge, or feedback so fast”
At its core, this project was built around three fundamental goals:
- Enhancing the user experience – making interactions with the Lending platform (BLP) more intuitive and efficient.
- Bringing valuable AI-driven insights to Fyndoo – exploring how artificial intelligence can streamline processes and improve information retrieval.
- Providing a real-world challenge for talented students – enabling hands-on learning, experimentation, and skill development in an industry setting.
“The most interesting part was AI. I had the opportunity to research and expand my knowledge on how to develop an AI Assistant using current technologies. AI is still quite new, and there are still many ways to implement it. It is similar to cooking—one dish can have multiple recipes that you haven't tried yet.” - Cindy Ye
An AI-powered chatbot has the potential to simplify the configuration process within BLP, making it more accessible and intuitive. It enhances the information retrieval process, ensuring users receive faster and more accurate responses to their queries. Beyond improving efficiency, this project provided Fyndoo’s product teams with a tangible AI use case—demonstrating how intelligent automation can streamline workflows.
This project was more than just an experiment—it laid the foundation for AI-driven advancements within Fyndoo. By uncovering both opportunities and constraints, it delivered key insights that will shape future innovation and development.
“Cindy not only proved that this was possible but also showed us how to implement it—and just as importantly, what its limitations were.”