This session is about the new problems that developers can encounter with new LLM-based projects. Some problems are easy to solve with a few experiments and framework switches. Others are more serious, like the difficulty of keeping context in memory, testing properly, forcing users to prompt correctly and finding the right balance between performance, cost, and complexity.

Key topics:
– Making the right choices with the design and components of modern AI apps – MCP + WAF approaches
– Developer testing problem and SDLC of the project
– Multi-agent/LLM approach for validation and secondary fact-checking.
– Finding a balance between RAG and plugins usage for different data sources, optimizing search
– Security gaps, lack of pentesting and intentional RAG poisoning

Join this talk to learn more about these problems, strategies to minimize their impact, and how to prepare for your next GenAI project.