The Future of LLM Reasoning: System 2 Thinking & Agentic Frameworks

LLMs are often treated as black boxes—users craft a single prompt and hope for the best. While prompt engineering, RAG, and fine-tuning have their place, they are not enough for complex reasoning tasks.
This session explores how to move beyond one-shot prompts by leveraging System 2 thinking, nuanced reasoning, and agentic frameworks.
We’ll cover:
- How LLMs mimic instinctive System 1 responses and why they struggle with deep reasoning.
- Techniques for breaking down complex problems into multi-step, structured workflows.
- Agentic frameworks that enable iterative refinement, context-sharing, and self-correction.
- Real-world use cases

Charlie Koster Bio:
I'm a Technology Leader passionate about elevating good software engineering practices. Areas of interest include architecture, distributed systems, agility, and functional languages.