As agentic AI is continuing to evolve, employers are looking to make sure they are keeping up with the latest updates.
ADP President, Global Product & Innovation Sreeni Kutam breaks down how AI agents are evolving in the workplace.
For more information, visit ADP.com/assist
Transcript:
Each of those tasks could be, by their own definition, agent. And let's focus and zoom in actually on agentic AI because you all have heard of that. And what is an agent? Let's take a look at the three, flavors of agentic AI.
An agent simply is combination of information and action coming together to do a task. So a simple agent, as a logical extension of that is you do a simple task by providing input information and connecting to an action through an API to get a single task done. So let me give you an example. You all probably hired somebody right? To work for you. The recruiter initially asks you, okay, okay, you have a opening, and I need to know the details about the job. You're trying to hide the person for. So you write a job description. You can go to any market facing, product today in general AI and ask the AI to write a job description. But what is important to to feed the information of your past company job description. So you have some disclaimers. You have some tone there as well as maybe we should look at the, job description of how other companies are doing and see if there are any best practices we can take it. All of that is information required for this task of writing a job description. So that's a good example to understand a simple agent.
However, an advanced agent is okay. What happens after you write the job description. You have to post it somewhere in the market so the candidates can apply. Once the candidates apply, you need to screen them. You have to schedule the interviews and then keep moving forward until you generate the offer letter. Each of those tasks could be, by their own definition agent an advanced agent. It sequences these simple agents together in such a way that end to end process can be done. So that's how to think about an advanced agent, a multi term. But there is always human in the middle to move it forward, one step at a time.
The autonomous agent is really the degree of the human in the loop, basically. So autonomous by nature is as you move forward in the sophistication of this agent journey, less and less independent. So that's how to think about.
