What are agents?
Agents are one of the hottest topics of generative AI these days. People talk about agents a lot, but there are many different definitions of what an agent is. LangChain itself defines an agent as “a system that uses an LLM to decide the control flow of an application.” While we feel it’s a great definition that is worth citing, it missed some aspects.
As Python developers, you might be familiar with duck typing to determine an object’s behavior by the so-called duck test: “If it walks like a duck and it quacks like a duck, then it must be a duck.” With that concept in mind, let’s describe some properties of an agent in the context of generative AI:
- Agents help a user solve complex non-deterministic tasks without being given an explicit algorithm on how to do it. Advanced agents can even act on behalf of a user.
- To solve a task, agents typically perform multiple steps and iterations. They reason (generate...
