First we had chatbots, then AI helping overworked support agents to improve customer experience. Now agentic AI (aka AI agents) is the CX hot topic, promising a focused and smart approach to addressing customer needs. But for non-tech types, what are they? And how do your business, and customers, benefit?
According to Gartner, AI agents are “are autonomous or semiautonomous software entities that use AI techniques to perceive, make decisions, take actions and achieve goals in their digital or physical environments.” Popular examples include OpenAI’s Operator and within enterprise tools like Oracle’s Fusion Cloud Human Capital Management (HCM) platform.
The boom in AI agents across sales, customer support and other departments means we should all pay attention to them. Recent research suggests a market worth $47 billion by 2030, but with the dramatic growth in interest, that could be a low estimate.
AI agents represent the latest form of old chatbots and AI services, working together to solve problems. They are “skills” that sit above your dumb data and tools. They use AI to think and make sense of that data. Agents then take action, providing answers to questions, suggesting solutions and actually performing tasks for you.
They are easy for businesses to create, and for those without IT skills, are available as pre-existing examples that can be launched in minutes.
How do you work with AI agents?
If you’re unsure where AI agents fit within your business, a handy explainer from Kustomer highlights four key areas you need to understand before adopting and deploying them.
They require you to define the core use cases that your AI agents will solve. Know your audience and the entry points they will use to interact with AI Agents. Optimise and structure your knowledge base, and understand how tools and actions work together.
For consumers or customers, AI agents are popping up across brands, services and within systems. The key difference to previous types of smart assistance is they can take a joined-up approach to any question.
For a example, when you used to ask “Siri to book a restaurant” it would show a list of locations on a map, leaving you to do the rest. The AI agent’s approach is to know your favourite places to eat, ask for the party size, make the booking with their app and add it to your calendar.
For customer service-facing apps, if there’s a solution to a problem, it can provide the details. If not, the agent will escalate the problem or make a repair booking depending on the situation.
As autonomous creations, agents can learn from their data environment, questions asked, and successful and unsuccessful resolutions.
Types of agent; reflexive, learning and goal-based
Since many businesses have different use cases and aims, there are several different types of AI agent available.
Reflex agents respond to what is asked of them. The simplest types follow a simple logic tree or workflow to reach a conclusion, much like a chatbot. Model-based reflex agents have more information and can learn from experience, asking for or adding insight to help resolve an issue.
Learning agents use feedback mechanisms to learn about their environment and get better at solving a question or respond to broader questions. They can refine how they work through information as they learn it.
Goal-based agents are given a target and use their knowledge or research to reach it. These are ideal for business analytics, as users ask how to reach a sales goal or design a more valuable product. They can plan and adjust their actions dynamically to meet those goals.
Utility agents consider the various options that are provided with and choose one that offers the greatest value or utility. The most common real world example is in autonomous vehicles and other critical systems.
With the basics understood, you can now identify the AI agents that will add value to your business. And, be ready as they get smarter across the rest of the decade to become a invaluable tool for all organisations.