Transforming Model UN Preparation with AI: The Vision of Karl-Gustav Kallasmaa

By Sheng Alferez

Karl-Gustav Kallasmaa, a distinguished engineer and Co-founder/CTO of Model Diplomat, played a critical role in transforming a previously non-functional AI platform into one of the most advanced political AI systems in the world. When Karl-Gustav joined, the AI chatbot was practically unusable, with response delays of up to two minutes just to display the first letter. He re-engineered the entire system to handle high operational scale and concurrency, enabling the platform to support upwards of 5,000 concurrent users with response times under three seconds. This leap in performance not only improves scalability but also delivers a smooth user experience even under heavy simultaneous demand, allowing students to rely on quick, reliable AI assistance for their Model United Nations (MUN) preparation.

What is Model Diplomat

Founded in 2023 by Georgina Songhurst, Model Diplomat is an AI-powered learning platform that helps students excel at Model United Nations conferences. With over 80,000 registered users, it streamlines research, speech drafting, and debate practice to enable users to prepare up to ten times faster than traditional methods. The platform offers instant AI-sourced insights into global policies and country positions, allowing students to research any country’s stance within five minutes, generate position papers in thirty minutes, and practice debates against realistic AI delegates. The system’s access to verified sources and feedback features enhances students’ confidence and performance, making it an indispensable tool for MUN preparation. Karl-Gustav’s infrastructure improvements ensure the platform can handle increasing user engagement while maintaining ultra-fast response times as the team continues to innovate and refine monetization strategies.

Technical Mastery and Innovations

Karl-Gustav designed and implemented a pioneering multi-LLM (large language model) orchestration system that coordinates several AI models to handle more than 800 political query types. This diversity—spanning breaking news to in-depth strategic policies—is 40 times greater than that of any competitor in the field, providing highly specialized, contextually accurate assistance tailored specifically for MUN preparation. Unlike generic AI assistants like ChatGPT, Model Diplomat uses precise prompts and fully analyzes source content, rather than relying on headlines alone. This specialized approach reduces what once required 20 hours of student research to just two hours of interactive AI use monthly, processing over 50,000 complex queries without downtime and ingesting more than 100,000 documents daily through custom pipelines.

Sophisticated Multi-Agent Intelligence Pipeline

The platform features an advanced multi-agent AI pipeline capable of operating in parallel under high demand. Key components include:

• An intent classification module that categorizes queries into over 800 types within 20 milliseconds.
• Dynamic source retrieval that simultaneously queries trusted providers such as Reuters, Financial Times, United Nations digital libraries, and more than 15 government APIs.
• An analysis engine that synthesizes conflicting information using credibility-weighted scoring.
• A fact validation module that cross-references claims against authoritative sources to ensure accuracy.
• A response synthesizer that generates coherent, citation-backed answers in real-time, usually within 10 seconds.

This sophisticated pipeline rivals top commercial AI platforms such as Anthropic Claude, Perplexity, and OpenAI, positioning Model Diplomat as a leading AI-driven edtech innovation.

Breakthrough Scale and Data Integration

Model Diplomat supports usage spikes with over 5,000 simultaneous active users from 73 countries, maintaining zero downtime regardless of demand intensity. Karl-Gustav’s breakthrough lies in seamlessly integrating multiple authoritative real-time data streams through custom-built adapters, unifying diverse APIs into a single knowledge graph. This institutional-grade news feed integration transcends earlier platforms’ limitations that relied on a single data source (like only UN Digital Library), providing students with a holistic and up-to-date research base essential for diplomatic simulations.

Engineering Excellence Underpinning MUN Preparation

Karl-Gustav’s expertise in distributed systems, real-time processing, multi-model AI orchestration, and enterprise data integration drove these innovations. His technical leadership expanded educational access globally and reshaped how AI can amplify human diplomatic capabilities rather than replace them. Rooted in his Computer Science background and thesis on explainable AI, Karl-Gustav applies theoretical rigor to practical challenges fueled by his passion for international relations. His commitment to diplomacy is further demonstrated by his involvement with the International Relations Society at the University of Tartu, where he organized a Model United Nations with 60 participants and led a team of 10.