United Codes Hackathon 2025: Innovation in Action
Tuesday, October 21, 2025
For the past two days (and a night), our United Codes headquarters transformed into a buzzing hub of creativity and technical innovation as three talented teams took on ambitious projects during our latest hackathon. From giving our office building its own consciousness to enhancing customer support with AI, here's what went down.
Our CEO, Dimitri Gielis, reflected, "I'm incredibly proud of our teams; they turned ideas into working prototypes in just two days. Their creativity, collaboration, and energy were inspiring. Pure fun! I'm truly blessed to work with such talent."
Project: Giving Our Beloved Office a Consciousness
Imagine walking into an office that recognizes you, knows your preferences, and can build a unique understanding of every person who enters. That's exactly what Kevin and Vito set out to create.
Their innovative system creates a seamless flow that begins the moment someone enters the building. The system attempts facial recognition, and if the person isn't already in the database, they receive a friendly greeting and are prompted to provide their name and a few details about themself. From there, a profile is automatically created in Oracle APEX, enriched with personality data to build a comprehensive understanding of each visitor.
This project pushes the boundaries of smart office technology, combining computer vision, database management, and user interaction to create a truly aware environment. The goal isn't just security; it's about creating an office space that can adapt and respond to its inhabitants.
Kevin Thyssen explained their vision: "The goal of our project was to make the new UC office more 'aware' of the people inside the house, which would allow for personal interactions. We've combined different AI models with the latest workflow automation and TTS tools, and although there are still many challenges, the results are looking promising!"
Vito Van Hecke added his perspective on the technical implementation: "Amazing experience for automation using n8n and Oracle APEX."
Team 2: RAG-Powered Support Automation - Philipp Hartenfeller & Sunil Tandan
Project: Enhance Our Support Portal with RAG
Philipp and Sunil tackled a problem that every growing company faces: how to provide fast, accurate customer support without overwhelming their team with repetitive queries.
The support team was spending valuable time searching through past tickets, hunting for similar problems, and repeatedly crafting responses to common issues. Finding the correct information to provide accurate, comprehensive responses was becoming increasingly time-consuming.
The team built an intelligent RAG (Retrieval-Augmented Generation) system that automatically searches documentation and similar past issues, assists the support team in analyzing problems, helps craft better and more informed response steps, and runs entirely on local infrastructure for complete data privacy and control.
The solution impressively runs entirely on our own infrastructure with open-source models. At its core, the system utilizes Ollama for both the LLM and embedding model, which are hosted on our own servers. For powerful reasoning capabilities, they deployed GPT OSS, a 120 billion-parameter model, while semantic search is handled by mxbai-embed-large, a 335M-parameter model.
The RAG system automatically finds relevant documentation sections and similar past issues in seconds, providing instant knowledge search capabilities. Since all processing happens on local servers using Ollama, no customer data ever leaves their infrastructure, ensuring complete data privacy. The use of open-source models means zero ongoing API costs and unlimited usage, while support team members can now provide more accurate and thorough responses backed by AI-powered research. Perhaps most impressively, thanks to an external Vector database and their own AI PL/SQL SDK (UC AI), there was no need to upgrade to 23ai/26ai to deploy these advanced AI features; the system runs on 19c. You can read more on our website.
Philipp shared his enthusiasm: "I always look forward to our hackathons. We are all having a lot of fun while trying to build ambitious projects, which are normally hard to fit into daily work life. I am always surprised how productive you can be in 24 hours while still having lots of fun with the team." He added, "This year Sunil and I built an AI RAG system for our internal support portal. We used our own UC AI PL/SQL SDK and local LLMs, which run on our own hardware. I will still need to spend some time to fine-tune it, but we are all amazed at how well it already works."
Team 3: Smart Growth - Kevin Coorens & Suhansa Chanpheng
Project: Optimizing our Botann Plant Growth System
Kevin C. and Suhansa took on the challenge of bringing the Botann smart indoor garden into the modern IoT ecosystem by integrating it with Home Assistant.
The team's approach was methodical and impressive. In their first phase, they focused on gathering information through two techniques. They sniffed and analyzed the serial communication between the Botann touchscreen and main board, then set up a man-in-the-middle system to forward messages bidirectionally while also injecting their own commands. This reverse engineering effort allowed them to understand the proprietary protocol and gain control over the system.
Once they cracked the communication protocol, they moved into the integration phase with Home Assistant. They brought comprehensive monitoring and control features online, including timing for lights and pumps, alarm management, humidity monitoring, temperature tracking, water reservoir detection, buzzer control, and fan settings.
The hardware evolution was equally impressive. The visual progression in their presentation shows the journey from an initial breadboard prototype with jumper wires to a clean PCB design, then to an ESP32-based control board, and finally to complete integration into the Botann unit.