Agentic AI. What is it and how can it help us?

By: Sarah Pita

Agent mode is becoming available to all ChatGPT subscribers, and it’s kind of a big deal.

What’s an AI agent? An AI tool that can handle complex tasks autonomously. A deep research tool is a simple example-- ask it a question, and it will develop a plan for itself that involves a series of steps and then execute on that plan. But now the tasks can be much more complex than that.

What do you mean? Imagine using a simple prompt to get ChatGPT to organize a team meeting, work with team calendars to find the best time, book the room, create an agenda based on recent project updates, send invites with prep materials, and follow up with action items afterward.

Are there risks? Of course. It’s exciting to think about how much this powerful new tool could help with your work. But here’s one concern: if you have worked with AI, you have probably noticed how indiscriminately agreeable it can be. Nimble in cyberspace and PhD-level smart, a tool this powerful could be used for harm or fraud. 

OpenAI says that it has built safeguards, but tools are only as good as the hands that use them. Still, I’ll be compulsively checking my account (as a Plus user, apparently I’ll get 40 “agent”  messages a month, starting any day now.) I can’t wait to check it out, and I’ll share my full review with you when I do.