Agents With Seatbelts: Practical Ways to Keep AI Code Gen Under Control, Jonathan Vila López

This talk explores best practices for implementing MCP setups and AI agents in development workflows to enhance quality and security. The speaker discusses the potential pitfalls, such as confusing interactions and accidental data exposure, that can arise from using these tools. They present comparative reports on various large language model (LLM) capabilities, highlighting differences in stability, output consistency, and necessary guidance to prevent errors. Additionally, the talk addresses common issues that lead to minimal productivity gains when adopting AI tools, emphasizing practical habits developers can adopt to create more reliable and secure AI workflows, thereby maximizing the benefits of AI code generation.