Yannic Explains: Agentic Tool Use

In this video, Yannic Kilcher, PhD, co-founder and CTO at DeepJudge, explains why tool use is one of the biggest leaps in making large language models (LLMs) more powerful. By themselves, LLMs can only generate text—but when paired with tools, they can interact with the world: asking users clarifying questions, reading and writing files, browsing the web, or running searches. A common setup is retrieval-augmented generation (RAG), where the model queries a search engine, pulls in relevant data, and combines it with your question to give a grounded answer. This pairing of LLMs with external tools is a core part of agentic AI, and it’s what makes AI systems more reliable and useful in practice.

‍