What’s agentic AI?
And why are you suddenly hearing about it?
The idea of AI agents isn’t new. Michael Wooldridge’s book ‘The road to conscious machines’ describes the idea of agents arising in the 1990s as a “self-contained, autonomous entity, situated in some environment and carrying out some specific task on behalf of a user.”
So why are AI agents suddenly everywhere again? As AI models like ChatGPT and Claude get more capable, the new promise is that they’ll move beyond a chat interface and be able to take on more tasks autonomously. Imagine that you want to book a ticket to the theatre. An AI agent could search the web for plays to see, check your calendar to find out when you’re free, check your history to be sure you haven’t seen it already, and book the tickets — all from a text prompt.
Until now, it’s certainly been possible to automate buying tickets. But it would have needed an engineer to list out all the steps needed and code up a specific ticket-buying workflow. For a one-off, that would have taken longer than just booking the ticket! Now, the idea is that AI models will be able to plan and execute the steps autonomously. And not just for booking tickets, but for a wide range of tasks.
That said, we’re still in the early days. Just as with LLMs, there are significant challenges to overcome, especially around security, accuracy, and accountability.
What do you think? Are AI agents the next big thing, or just another passing trend?
I work with companies building AI technology. Get in touch to explore how we could work together.
