When an AI system produces poor results, the model is often the first thing blamed. In many cases, however, the issue lies in the context surrounding the model rather than the model itself.

This session introduces Context Engineering, the practice of designing the information environment in which AI systems operate. We will examine how instructions, retrieval, memory, formatting, constraints, and workflow decomposition influence model behaviour, and why these factors often have a greater impact on outcomes than many teams expect.

Through practical examples, attendees will explore techniques for improving the quality and reliability of AI workflows by refining the context provided to the model. The session focuses on approaches that can be applied without fine-tuning models or building additional infrastructure, demonstrating how changes to context can significantly affect system performance.

By the end of the talk, participants will have a clearer understanding of why AI systems sometimes fail to produce expected results and how context design can be used to improve those outcomes.

What You Will Learn:

• How instructions, retrieval, memory, formatting, constraints, and workflow decomposition influence model behaviour
• Practical techniques for improving AI workflow quality through better context design
• Why refining context can often improve outcomes without changing the underlying model

Who Should Attend:

• AI engineers
• Software engineers building AI applications
• Application developers working with LLMs
• Technical leads
• Architects designing AI systems
• Teams responsible for AI workflow design and implementation