Want to boost customer satisfaction? AI agents are the future! Automate tedious tasks & provide personalized experiences. #AgenticAI #CustomerExperience #GenAI
Explanation in video
Hey everyone, John here! You know, AI is popping up everywhere, and one area it’s set to really shake up is how businesses help you, their customers. We’re talking about something called “agentic AI,” and today, we’re going to break down what that means for your future customer experiences. It’s all about making things smoother, faster, and maybe even a little bit magical!
So, What’s This “Agentic AI” All About?
Imagine you have a super-smart personal assistant, but instead of a person, it’s a clever computer program. That’s kind of what an AI agent is! These aren’t your old-school, clunky chatbots that can only answer a few pre-programmed questions. We’re talking about AI that can understand what you need, figure out how to do it, and then actually take action.
Lila: “John, that sounds a bit like science fiction! You said ‘agentic AI.’ Is that different from the AI we hear about that creates art or writes things?”
John: “Great question, Lila! You’re thinking of generative AI (genAI), which is fantastic at creating new content. Agentic AI often uses that genAI power, but it takes it a step further. An ‘agent’ is something that acts or does things. So, agentic AI isn’t just about generating a response; it’s about being an active helper that can perform tasks, make decisions (within limits, of course!), and manage complex interactions. Think of genAI as the brain, and agentic AI as the brain and the hands, working together to get stuff done.”
Businesses are realizing that instead of making you click through endless menus or fill out long forms, they could use these AI agents to understand your preferences and help you out much more easily. The big idea is that soon, these AI helpers will be the standard way we interact with companies.
John Kim, who’s the CEO of a company called Sendbird, puts it nicely. He says businesses will start using specialized AI agents that are experts in specific things like product details, stock levels, pricing, or even the legal stuff. He even thinks that in the future, we’ll all have our own personal AI assistants for different parts of our lives, like one for managing our money, another for entertainment, and so on. Pretty cool, huh?
Starting Small: Let AI Handle the Annoying Bits First
Now, some companies that jumped into AI for customer service a bit too quickly ran into some problems. It’s new technology, after all! So, many businesses are being a bit more careful this time. They’re focusing on important groundwork like making sure the AI is used responsibly (that’s AI governance – like setting rules for the AI), ensuring the information it uses is accurate (data quality), and, of course, lots and lots of testing.
Lila: “So, they’re not just going to let AI run wild with my bank account details, right?”
John: “Exactly, Lila! Safety and accuracy are super important. That’s why a smart way to start is by using these AI agents for the really boring, repetitive tasks. Think about the stuff that frustrates you as a customer and happens all the time.”
Dave Singer from a company called Verint suggests that AI can be a champ at things like:
- Asking the right initial questions to understand what a customer needs.
- Quickly searching for answers to customer questions.
- Handling the paperwork or notes after a customer service call (they call this “post-call wrap-up”).
By letting specialized AI bots handle these “micro-workflows,” human customer service agents have more time for complex problems. This means better customer experiences and can even save companies money or help them make more.
Making Information Easier to Find
Another good starting point is product information. Instead of making you dig through pages and pages of instruction manuals or FAQs, an AI agent could just answer your questions directly. Imagine you’re trying to set up a new gadget. Wouldn’t it be great to just ask an AI, “How do I connect this to my Wi-Fi?” and get a simple, clear answer?
Jon Kennedy from Quickbase suggests companies should look at where customers usually go for help – like help pages or online forums – and think about how AI can make those better. He also says AI can help guide customers through their “journey” with a product, pointing them to the next helpful step.
Deon Nicholas from Forethought takes it a step further. He says while it’s great for AI to find information quickly (he mentions something called RAG-based search, which we can chat about!), it’s even better if the AI can do things for you.
Lila: “Hold on, John! ‘RAG-based search’? That sounds like something my grandma uses for cleaning!”
John: “Haha, not quite, Lila! RAG stands for Retrieval-Augmented Generation. It’s a bit of a mouthful! But basically, it means the AI doesn’t just pull answers out of thin air. First, it ‘retrieves’ or finds relevant information from a specific, trusted source – like a company’s own help documents. Then, it uses that information to ‘augment’ or improve the answer it ‘generates’ for you. So, it’s like the AI does a quick fact-check before it speaks, making its answers more accurate and reliable.”
So, instead of just telling you how to reset your password, an AI agent could actually help you reset it or check your order status directly. That’s a big step up!
Good AI Needs Good Information (Like a Chef Needs Good Ingredients!)
For AI to really shine in customer interactions, it needs access to good, clean, and well-organized data. Think of data as the food an AI eats. If it eats junk food (messy, incorrect data), it won’t perform very well!
Companies use special tools to gather and organize all their customer information.
Lila: “You mentioned tools like ‘customer data platforms’ and ‘data fabrics’ in your notes, John. Are those like super-smart filing cabinets?”
John: “That’s a great way to think about it, Lila! A customer data platform (CDP) is like a central hub that collects all the information a company has about its customers from different places (website visits, purchases, support calls) and puts it all together in one neat, organized spot. This gives the company a single, clear picture of each customer. A data fabric is a bit more like a clever network that connects different data ‘silos’ or storage spots without having to move all the data to one place. It allows different systems and AI to access the data they need, wherever it lives, in a secure and efficient way. Both help make sure the AI has the right information to be helpful.”
Tara DeZao from Pega emphasizes that a good AI customer experience strategy depends entirely on good data and rules for how it’s used. And, super important, companies need to keep testing and learning to make sure the data stays fresh and accurate. This not only helps the AI do its job better but also protects the company’s reputation.
Keeping Your Data Safe is Key
When companies bring all this customer data together, they absolutely must have strong controls for security, who gets to see what, and making sure users are who they say they are. Many use something called data security posture management (DSPM) platforms.
Lila: “DSPM? Another one of those techy acronyms, John?”
John: “You got it, Lila! DSPM stands for Data Security Posture Management. Think of it as a high-tech security guard for a company’s data. It helps businesses understand where all their sensitive data is, who has access to it, how it’s being protected, and if there are any security weaknesses. It’s all about keeping customer data safe, especially when it’s stored in lots of different places, like in the cloud.”
Experts like Osmar Olivo from Inrupt suggest that companies should move towards data systems that are centered around the user, allowing for more personalized and responsive experiences. He also says it’s important to let users give feedback to help the AI learn and get better, and even correct it if it gets something wrong.
But getting data right is tricky. Manish Rai from SnapLogic warns that a huge number of AI projects (he estimates over 80%!) can fail because of problems with connecting to data, poor data quality, or people not trusting the data. Success comes from tools that make it easier to build these AI agents, get data ready for AI, and keep an eye on things to ensure accuracy.
Sometimes, as Rosaria Silipo from KNIME points out, it’s good to have a “human-in-the-loop” – a real person who checks the AI’s work, especially for important tasks. Or, you can even have other “guardian” AI agents whose job is to check the results of the main AI!
Taking Customer Service Calls and Chats to the Next Level
We’ve all had frustrating experiences with customer service, right? Waiting on hold, explaining your problem over and over… Well, AI agents aim to change that. One survey even found that 23% of people would rather watch paint dry than deal with bad customer service repeatedly! Ouch.
Instead of basic chatbots that follow rigid rules, new AI agents can sift through information and respond to customers much more intelligently. This frees up human agents to handle the really tricky or sensitive cases, often with an AI assistant helping them out too.
Vinod Muthukrishnan from Cisco says there’s a clear link between customer happiness and good self-service options. He believes these advanced AI agents can manage the entire interaction between a customer and a brand, offering smart, seamless help whenever and wherever the customer needs it.
Part of the challenge is not just the data, but also that many current customer service systems were built in pieces, only addressing one part of the customer’s journey. Technologists are now using approaches like design thinking.
Lila: “John, what’s ‘design thinking’? Does it mean making things look pretty?”
John: “Looking pretty can be part of it, Lila, but design thinking is much more than that. It’s a problem-solving approach that puts the human user – in this case, the customer – at the very center. Instead of just adding new tech to old systems, design thinking encourages businesses to deeply understand the customer’s needs, frustrations, and goals. Then, they design (or redesign) the entire experience from the customer’s point of view to make it as easy, effective, and pleasant as possible. It’s about building solutions for people, with people.”
Chris Arnold from ASAPP points out that using a powerful AI like an LLM (Large Language Model – the engines behind many genAI tools) to create a personalized, conversational experience is far better than the often clunky, transactional experiences we get from many apps and websites today.
Super Important: Test, Test, and Test Again!
Before companies let these AI agents loose to help customers, they absolutely need to test them thoroughly. Miles Ward, the CTO of Sada, says you can’t just “throw an agent out there untested and unmonitored.” That would just create new problems!
There are basic safety measures, like filters to stop the AI from saying inappropriate things or going off-topic. But brands need to go further and make sure the AI agents are:
- Accurate: Is the information correct?
- Appropriate: Is it responding in a suitable way?
- Ethical: Is it fair and unbiased?
Ganesh Sankaralingam from LatentView suggests testing AI responses across five key areas:
- Relevance: Does the answer actually relate to the question asked? (Like, if you ask about shoes, it doesn’t talk about cheese!)
- Groundedness: Is the answer based on the facts and information it was given? (It’s not just making stuff up.)
- Similarity: How close is the AI’s answer to what a perfect human answer would be?
- Coherence: Does the answer flow logically and make sense, like how a person would talk?
- Fluency: Is the language good? Is it grammatically correct and easy to understand?
Deon Nicholas (remember him from Forethought?) adds that businesses should test AI by seeing how it handles past customer questions, measure how often it can solve issues all by itself, and even use another AI model to check if the conversations are positive and accurate.
What’s Next? The Future of AI Agents in Customer Service
So, where is all this heading? Mo Cherif from Sitecore advises companies not just to add AI to what they already do, but to completely rethink the customer journey with AI in mind from the very start.
There are a couple of different views on how these AI agents will evolve:
- More Autonomous AI: Some experts believe AI agents will become more independent, able to make complex decisions and handle a wider range of tasks on their own. Michael Wallace from Amazon Web Services gives an example: imagine an airline’s contact center facing a storm. Agentic AI could automatically rebook passengers and send out notifications, all without human intervention, letting human staff focus on the most complex customer needs.
- More Human-Centric AI: Others think AI will be more like a partner to humans. Doug Gilbert from Sutherland Global argues that AI isn’t about replacing people, but about making interactions smarter, faster, and more natural. The secret, he says, is AI that learns from real-world interactions and becomes more intuitive and less robotic over time.
It’s likely we’ll see both types of AI agents in the future. Some tasks will be fully automated, while for others, AI will work hand-in-hand with human experts.
Our Quick Thoughts
John: “You know, this is pretty exciting stuff. The idea of AI making customer service less of a headache is something I think we can all get behind. It’s clear that getting the data right and testing thoroughly are the big hurdles, but the potential to make things genuinely better for customers is huge. I’m cautiously optimistic!”
Lila: “I agree, John! As someone who isn’t a tech expert, the thought of AI assistants that actually help and don’t just confuse me more sounds amazing. I like the idea of starting with the boring tasks – if AI can handle those, then the human experts can help with the really tricky problems. I just hope companies remember to keep it simple and user-friendly for people like me!”
The bottom line for now is that businesses need to really understand what their customers need, work hard on getting their data in top shape, and be super diligent about testing. It’s a journey, but a promising one!
This article is based on the following original source, summarized from the author’s perspective:
Design a better customer experience with agentic AI