Building a Retrieval Augmented Generation Bot in Just Seven Hours

The Pre-Game: Competitive by Nature

Thursday finally arrived. My team and I were battle-ready, totally fueled by caffeine. Joe brought cupcakes to provide the necessary sugar rush; Daryl was sporting his new glasses, determined not to miss a single semicolon; Adnan was already deep into Terraform and Lambda while Michal mapped out the UI/UX before the first coffee was even finished. As for me? I was ready with my repertoire of "ugly faces" to break the concentration of our rivals.

To be honest, I’ve always been extremely competitive. It probably started with Pong on my first Atari. In my daily life, my legs seem to compete to see which one reaches the office first, and my internal operating system doesn’t have a "power-save" mode—only "execution" and "competition."

Since I joined the tech world, the idea of a Hackathon had been lingering in my mind as the ultimate challenge. When Dot Collective announced this event, I didn’t just see an opportunity to experiment with cutting-edge tech; I saw the perfect stage to show what we’re capable of and, perhaps, finally clear my CV of that streak of second-place finishes.

The Challenge: An Enterprise-Grade AI Lab

The organisation was truly exquisite. The platform was ready for us from minute one, allowing us to hit the ground running and focus entirely on the build. The presence and support of the "gremlin hunters" was absolutely stellar—having that expert backup to resolve technical glitches in real-time made the whole experience seamless and kept our momentum high.

Our goal was to build a Retrieval Augmented Generation (RAG) bot in just seven hours. The mission was to transform our internal company handbooks and policies, traditionally stored in Confluence, into an interactive and intelligent knowledge base. We set out to build an enterprise-grade solution—one that aligns with corporate values, understands its limitations, and remains secure, precise, and professional; allowing any Dot Collective employee to query the handbook with natural language, simply asking: "How many days of paternity leave do I have?" and getting an accurate answer instantly.

The focus was on creating a production-ready blueprint for our clients, built within a corporate cloud environment. For this challenge, that meant an AWS-managed setup using Bedrock, OpenSearch, Lambda, and the latest LLM models. This experience will serve as our high-speed laboratory—a space to push AI to its limits and extract every possible advantage for our clients.

The Breakthrough: Synergy Over Rivalry

As the clock ticked down, my "win-at-all-costs" mentality shifted into something far more valuable: team synergy. Watching Adnan hitting milestones with Terraform and Lambda, Michal managing the UI/UX with expertise, Joe configuring the Guardrails with surgical precision, and Daryl stress-testing the model while crafting a killer presentation reminded me why we love what we do. The energy was contagious.

(I did my part too, of course!)

At the end of the day, the competitive outcome was almost secondary. What we truly took home was accelerated learning. There’s something about the mix of learning and adrenaline that sears concepts into your memory like nothing else. We didn't just build a bot; we built a blueprint that we are now ready to pitch to our clients with the confidence of those who have tested the tech in the trenches.

My Final Thoughts

Dot Collective didn’t just organise a contest; they created a platform for us to connect with the future of AI. By testing these boundaries in an intense, real-world scenario, we ensure that when we deliver a solution, we’ve already cleared the hurdles and maximised the technology's potential.

I went home with much more than just a great experience; I left with the certainty that while personal ambition makes you fast, team talent makes you unstoppable. This experience, and the hackathons to come, are now an indelible part of our journey.

And let’s be real…Joe’s cupcakes were definitely a deciding factor in our success.

P.S. Wait, did I mention?

We won!!