What Are Software Agents?
The idea of being able to use agents to actually solve real problems for us is one that’s still evolving in the heads of most people. So, what is a software agent? Well, first off, we pretty much understand what conversational agents are, right? When we look at conversational agents, you talk to some software system, and it will send somebody off to do a task for you — maybe transfer money, pull up a document, or do a search. We all know that. But those are simply transactional agents, like performers of transactions. However, they don’t do anything a whole lot more sophisticated than that.
How AI Agents Are Being Used Today
Most people aren’t aware that agents can do things that are long-running. They can actually have memory, they can have a state, and they can stay around to do very complicated processes for us. But because we're not really familiar with what we do with such a capability, it’s very difficult to tell the providers exactly what we're willing to pay for.
As I said, it’s very difficult to know what you should do with an agent. Most people don’t even really know what agents are. In simple terms, an agent is a piece of software that will execute tasks for you. That’s the simplest way I could put that. It's the type of software that's going to be able to execute simple tasks. It might be transactional or it might be able to do things that are more complex, where it actually talks to other systems — maybe other agents even — to get the job done.
We have these systems of agents or multiagent systems. You can get into some very interesting scenarios where agents can do things that human beings might never even consider doing.