An Answer to Alex’s Problem: DialogueAI
Written by Dialogue AI
How much can you truly change in just 12 hours? Can you even scratch the surface of a problem meaningful enough to matter? That was the question hanging in the air, thick with the combined intellectual energy of the bright young minds from over 43 universities. The pressure was immense, a silent challenge to innovate, to create, to leave a mark. "If you wanna make the world a better place, take a look at yourself and make a change." But for us, the answer wasn't in the mirror; it was sitting right across the breakfast table. It was our friend, Alex.
With over 150 participants ready to battle it out in a 12-hour coding marathon, the stakes were high. Yet, amidst the clatter of cutlery and the nervous pre-Hackathon chatter, a new negotiation was taking place. We were half-jokingly discussing the need for a lawyer on our team when Alex, with a grin, began to seriously negotiate his own position. We hadn't even made it out the door, and a decision had been made: Alex was in. It wasn't just his legal mind we needed; it was his problem.
Alex, a law intern at a prestigious private equity firm, confessed his biggest struggle: a profound lack of mentorship. He was navigating high-stakes law, learning to swim by himself. It was a story of ambition meeting isolation. What if, we pondered, we could build him what he was missing? What if we could create his own personal senior lawyer, an ever-present mentor, a tireless researcher? That moment marked the birth of "Dialogue AI".
The day before the hacking began, the organisers hosted a “Tech Showcase” where every platform offering tools and services for the Hackathon set out exciting demos and live walk-throughs. RegGenome’s team unveiled their regulatory-data API; Mistral AI delivered a session on building secure, private and customisable AI; Groq engineers showed off lightning-fast inference with the Groq API; and the founder of Momen demoed a no-code builder for streamlining repetitive cognitive tasks. But when the Google Cloud engineers walked onstage and started chaining their comprehensive services together, the decision became crystal clear.
Watching them, we could see our own architecture materialise in front of us: three specialised AI agents powered by Gemini, orchestrated through Firebase Extensions, with Vertex AI’s vector store handling embeddings for our Retrieval-Augmented Generation (RAG) implementations. By the end of the session, we weren’t merely impressed—we were convinced. The decision definitely paid off as we later won the Google Cloud Runner-Up Prize for our inventive use of their platform and suite of tools.
The Hackathon day itself was a 12-hour blur of controlled chaos and intense focus. We made use of the different rooms offered by Judge Business School to put our heads together and get to work. During the hacking, there were several ongoing lightning talks being held on various topics relating to Law and AI. We took turns attending these insightful ten-minute sessions, hearing leaders from major LegalTech firms describe how they’re reshaping jurisprudence with code was equal parts eye-opening and reinvigorating; each talk sent us back to our laptops with sharper focus, a fresh bundle of ideas, and renewed conviction that we were hacking on the frontier of a rapidly changing industry.
Throughout this whirlwind of development, Alex became our "lab human". With every new feature we deployed, he was the first to test it, his initial skepticism slowly melting away, replaced by a smile that grew wider and wider with each iteration. We were witnessing the birth of a solution in real-time.
Our creation, Dialogue AI, was more than just a tool; it was a revolution in legal training. It allowed Alex to input the intricate details of his case: the presiding judge, the opposing counsel, and the nuanced facts. From there, a trio of AI agents would engage in a dynamic debate, relentlessly challenging each other to unearth the most optimal arguments. The system dove deep, researching the personas involved—analysing the judge's preferences, dissecting the opposing lawyer's negotiation tactics, and even identifying patterns in their past victories and defeats. It then presented Alex with a comprehensive arsenal of resources for a deeper, manual dive.
But we didn't stop there. We took every piece of data—every persona, every case detail, every subtle nuance—and engineered a hyper-realistic simulation. In this virtual courtroom, the opposing AI lawyer didn't just argue; it embodied the very essence, style, and persona of the real-world opposing counsel. Here, Alex could stand and argue his case, receiving live, actionable feedback and profound insights into the strength of his arguments. At the end of each session, we provided a detailed summary of his strengths and weaknesses, empowering him to make his limited mentorship time laser-focused and incredibly effective.
Our first pitch felt electric. The judges leaned in and we saw their initial curiosity transforming into enthusiasm. We had not just identified a problem; we had presented a powerful, elegant solution. The grand finale became our next challenge; the competition now fiercer than ever. When the dust settled and the final announcements were made, it wasn't just a product that had secured second place at the Hackathon. It was our answer to Alex's problem. It was Dialogue AI.
Nivar Anwer is an Artificial intelligence graduate holding multiple AI researcher and Research assistant positions. His primary focus is on medical imaging, exploring the intersection of image processing, computer vision, Large Language Models (LLMs), and Vision-Language Models (VLMs) to improve the field.
Alina Ahmed is an Artificial Intelligence BSc graduate with strong academic foundations in machine learning, neural networks, and intelligent robotics. Having worked in software engineering roles, she is experienced in integrating AI into advanced technical projects and developing data-driven systems to solve real-world problems. She is passionate about building impactful solutions that can transform industries through AI and ML.
Alexander Plester is a German final-year law student at the University of Münster with practical experience in commercial law firms in Germany and Luxembourg. In addition to his strong interest in corporate law, he is convinced that legal technology and AI are transforming not only the legal profession but also access to justice. With a background in international law, he is particularly focused on the automation of cross-border legal workflow to make complex processes more efficient and scalable.
Valentin Colomer is an engineering student at IMT Atlantique in Brest, France, specializing in Software Engineering. He is currently working for a French company and plans to focus on state-of-the-art distributed software during his final year at IMT Atlantique. He holds a two-year degree (DUT) in Computer Science from the IUT of Lannion.
Jordan Baumard is a French student currently pursuing an MSc in Engineering at IMT Atlantique in Brest. He previously earned a BSc in Computer Science from IUT de Paris - Rives de Seine. Passionate about computer science and programming, Jordan thrives on solving complex problems and is always eager to explore new technologies. He has a strong interest in artificial intelligence and enjoys collaborating with peers to develop innovative solutions.