Agentic AI and Customer Success Managers (CSMs) can both drive significant value to your business. 

But here’s the question: Can they work together? And if so, what is the equation between them?

Look at your current workplace. How many of you work with a virtual assistant or a virtual agent? Many of you do. Now, think five years into the future. By then, practically all of us will be working alongside these digital entities. 

But before we can understand how we fit into this future, we need to understand what agentic AI actually is and how it differs from the AI we’ve used in the past.

In this article, you’ll discover the evolution of AI technology, see real-world examples of agentic AI in action (including a personal experiment involving cake), and we will conduct a "face-off" to see which skills belong to the human and which belong to the machine.

From prediction to autonomy

To understand where we are, we have to look at the roadmap of how we got here. A common misconception is that Generative AI came after Agentic AI. 

The truth is, Generative AI formed the groundwork on which Agentic AI was built.

Customer Support Reimagined with Generative AI
Transform your support strategy with generative AI. Learn how Crestron boosts CX, efficiency, and agent impact without losing the human touch.
Customer Success CollectiveCustomer Success Collective

Here is the actual timeline:

  • Predictive AI (~15 years ago): This was about understanding data, analyzing huge sets, forecasting trends, and suggesting actions. It was useful for forecasting and lead scoring.
  • LLMs (2018): Large Language Models transformed the horizon. This was the first time coding was done in natural language, bringing a human-like element to the form (think GPT-2 and GPT-3).
  • Generative AI (November 30, 2022): This is the day ChatGPT launched to the public. It took the power of Gen AI from behind closed doors and gave it to everyone. It can generate text, images, and even music based on data patterns.

Then, in 2024, came Agentic AI.

This is the next step. While Gen AI generates content based on data, agentic AI can autonomously take actions on that data. These agents don’t just analyze; they initiate.

Defining Agentic AI

So, what exactly is it? Agentic AI refers to artificial intelligence capable of autonomously pursuing goals, making decisions, and initiating actions based on context – without needing step-by-step instructions.

It gets even better: it can improve based on human feedback. If an agent takes an action and you tell it, "That was not what I wanted," it will learn from that feedback and adjust.

We are already seeing this in action across industries:

  • Service agents: Going beyond chatbots to actually resolve issues.
  • Sales development: Running sales cycles and generating leads.
  • Campaign assistants: managing marketing workflows.
  • Personal shoppers: Pandora, for example, uses an assistant to learn your style and make specific recommendations.

Case study: The cake agent

I want to share a personal story about building my own agent. When Salesforce launched Agentforce, I was given the chance to create a custom agent in a demo environment. While my colleagues were building serious sales and service agents, I am a baking enthusiast.