Multi-Agent in Action: Solving a Market Risk Breach Scenario

Most AI demos look impressive… until you try to deploy them in production.

That’s where things start breaking.

From hallucinated outputs to lack of traceability, today’s LLM applications struggle when exposed to real-world enterprise workflows. Especially in high-stakes domains like market risk, where every decision must be explainable, auditable, and compliant.

In this Live Problem Solving Session, we go beyond theory and walk through how to design a production-grade multi-agent system that can investigate a real market risk limit breach end-to-end.

What You’ll Learn

We’ll break down how to move from fragile AI systems → controlled, reliable architectures

You’ll learn how to:

• ✅ Decompose complex workflows into bounded agent responsibilities
• ✅ Separate data retrieval from reasoning
• ✅ Introduce a deterministic orchestration layer
• ✅ Add validation and confidence gating
• ✅ Build systems that are auditable and production-ready

Pre-Requisites

• ✔️ Understanding of LLMs
• ✔️ Familiarity with workflows or pipelines
• ✔️ Familiarity with LangGraph is recommended for better understanding of orchestration concepts
• ✔️ Curiosity to learn how real-world AI systems are built