Agentic systems with artificial intelligence (AI): emerging and disruptive technologies

Abstract

Introductory talk on the evolution of artificial intelligence up to agentic systems, with a focus on agronomy. It starts from the current context (2024 Nobel Prizes, definition of AI, enabling factors such as compute, data and algorithms), goes through the history of data analysis and the ChatGPT turning point, and explains neural network architectures (CNN, RNN/LSTM and Transformers) with agricultural applications. It then defines agentic systems, their types and architectures, ecosystems such as LangChain/LangGraph, n8n, Copilot Studio and OpenAI Assistants, and protocols such as MCP, A2A and AG-UI. It shows use cases (Cereza, Nutribot, Feedibot, SoilGrids, GenCast, Planet Parasol) and closes with resources.