Hype Cycle for Agentic AI: What Leaders Need to Know

Ajouté : 2026-06-03

[00:00:00]Welcome to Gartner Thinkcast. I'm Karen Stokes Lockheart. Today we're unpacking one of the most talked about topics in enterprise technology right now. Agentic AI. If it feels like you've been hearing that term everywhere lately, you're not alone. The hype is real, but so is the potential. To help separate signal from noise, Gartner has recently introduced a new hype cycle specifically for Aentic AI. Joining us today to help break it down is our own subject matter expert Gartner distinguished VP analyst Rajes Kandiswami. Rajes, welcome in.

[00:00:36]>> Thanks Karen. Good to be here.

[00:00:39]>> Wonderful. Well, let's jump right in and kind of start with the foundation for any listeners who may be a little less familiar. Can you tell us what exactly is a Gartner hype cycle and what is what is it designed to help leaders understand?

[00:00:53]>> Sure. Gartner hype cycle is one of the most popular Gartner deliverables. What it is, it really unpacks a particular technology, especially the ones that we hear a lot about and where hype is increasing. The hype cycle helps cut through that and then identify what are the different aspects of the technology and what they can deliver and really where the hype is compared to its potential.

[00:01:24]And so when you look at that hype cycle, how do you and your team actually determine where a technology sits on the hype cycle and maybe a little bit about what signals, data or patterns you're looking for?

[00:01:36]>> Yeah, sure. So Gartner, we have a varied set of analysts who and we are constantly talking to clients across all industries and technology companies. So what we do is we pay attention to the specific technologies that are increasing in attention and then a group of us spend time to look through to see >> in the technology what are the different components what are the different aspects of it.

[00:02:04]>> What are companies doing with it? Are there vendors that are offering interesting new innovations? We look at all of that together. We discuss what is real versus what is hype and then make a collective decision on what hall needs to be in the hype cycle, how big each of those transformations could be and where is the hype against how long it's going to take for it to really realize its potential. So it's an elaborate process.

[00:02:32]You looking at things in the market, talking to our clients and many of us discussing it internally.

[00:02:39]So, I know many people see the hype cycle as a prediction tool, but as you're talking, it sounds like it's really more about behavior and expectations.

[00:02:48]Um, like how should leaders interpret it and maybe just as importantly, how should they not interpret it?

[00:02:54]>> Yeah. The the way to use the hype cycle is, you know, for instance, in this case, agent TKI, you hear all the hype in the market.

[00:03:02]>> Yeah. And now what is important for leaders to do is really understand what are the different innovations that drive agent AI. So that is the first thing for leaders to understand and use the hype cycle for and second getting an understanding of of the different things that you want to use. Where is the hype and where is the potential and more importantly >> what is going to be ready within the technology to use in the short term and which ones might take some time to realize and using that you know rich set of information to plan your right investments and any architecture decisions >> and any use cases that you want to implement all of them. So what we ask is while there is overlap hypervoted technology go a level deeper use the hype cycle to find what are the specific components and where they are and then make a more judicious decision on your investments.

[00:04:00]>> So Gartner already covers AI broadly. So what makes aentic AI distinct enough to warrant its own dedicated hype cycle this year? Yeah, Karen, there is not a day that goes by when you don't hear about uh AI agents and agentic AI in the media. Within AI, this is the most hyped of technologies.

[00:04:23]And what we see is while the hype is there, the interest in CIOS wanting to invest is also very high. We did a survey last year with CIOS asking the about different emerging technologies and what they plan to invest in. agent KI is the number one place they wanted to invest in this year.

[00:04:43]Now >> while that's true a key point that we wanted to make with the height cycle it's just not about the hype it is the fact that agentic AI is not one technology there are a sort of technology innovations each of them that has got to have different type of impact that are progressing at different speeds and all of them collectively work together so we introduced you know a dozen or more technology profiles as part of this hype cycles that was not there before. So it is important for the audience to know all these different technology innovations and components that they or their team needs to master to succeed with agent AI and >> we also we can't ignore all the talk about AI agents taking people's jobs.

[00:05:32]What are we actually seeing play out with how companies are using agents today?

[00:05:37]>> Yeah, it's uh uh the the fear and the worry is real. No technology has captured people's imagination and no technology is close to what we do as humans at the workplace. So it is real.

[00:05:50]We cannot diminish those worries. We are still in early days. In the early days what we find is companies are identifying interesting ways to use AI agents that is not necessarily how people do their work. Are doing work that is not possible by a single human.

[00:06:09]For instance, processing significant amounts of data at rapid speeds, doing work that needs to get done for the companies to be more efficient or productive or do more but not a single human can do. That's what we see in these early days. But this technology is maturing and we as like you know in the business world are still trying to figure out how all to use AI agents and as we figure out this movie can like you know take different forms and shape over time >> looking even further out what does success with aentic AI actually look like within an enterprise >> yeah the you know right now many organizations are investing in AI agents and you know in pockets right and they're trying to scale it. But if you play this over a long period of time, we see AI agents become a crucial element for an enterprise to achieve its objectives because it promises like you know good quality work done at rapid speed at no cost.

[00:07:13]>> Now organizations will use it in a variety of ways. One in places that's of competitive advantage to you. How can we use AI agent to strengthen our competitive advantage? This is more likely going to be in places that are external facing. So how we engage with the customer, offer them a very specific product or service uh and doing all those things better. And it also could be internally trying to do things in unique ways that your competitor cannot do. So how do I use AI agents to elevate my competitive advantage?

[00:07:49]Second, how can I use AI agents where it is commoditized work? How can I use AI agents to do it done faster, cheaper, better without increasing my cost as I grow as an enterprise? That is another place that people will invest in. So in the future what we see is a use of AI agents proliferating across the enterprise on the business side functional side and in IT to achieve the these things that's what we see over time >> it is cutting across all aspects and there are some stats right it's so it's noted that fewer than 20% of organizations have deployed aentic AI but to your point about everyone wants to deploy it more than 60% plan too soon. Can you talk a little bit about what's behind the gap between interest and execution and how quickly do you think that gap will close?

[00:08:45]>> So Karen, when we spoke earlier, I mentioned about all the hype, >> but behind all the hype, >> the potential is real. Agentic AI can help do achieve a lot of things with AI that is just not possible with other aspects of AI and executives realize that across industries. So they are keen on applying AI agents. So what we find is agentic AI for all the hype and all the visibility today is fairly new. We have not had these new version AI agents that are driven by these large language models not even for 18 months. They're fairly new. So while when you talk to executives across industries while they are aware they're already busy with the real AI work that they've been doing today and that work is piling up. So it is going to take some time for them to spend time to build with this technology and further they need to identify places where they can really apply this technology to get value. So we are in very very uh early first steps of this long journey in agenda.

[00:09:59]>> Okay. And I know a lot of the conversation today is focused on building agents but you've talked about some of the real challenges you know lying in other places whether that's the governance guard rails security. Can you talk about why some of those areas are lagging behind and why they matter so much? Yeah, I think uh the first let me spend time on why they matter so much.

[00:10:24]So the key thing is agents are not applications. Agents can perceive what's occurring in the world. They can take actions on your behalf. They can make decisions or they can go after goals and they can do it like you know either by themselves autonomously or semi-autonomously. And we urge to leaders listening to this to think more about the semi-autonomous ones and keep some humans in the loop. Now be all these things that the agents do more than applications if not done properly they can also create a lot of harm. So it becomes vital that organizations have good governance. What type of agents can be built? What can they do and cannot do? These things need to be really thought about. And second in terms of what they are allowed to do and putting guard rails so that they don't end up hurting your reputation or your finances um or your relationship with your customers those things are important.

[00:11:27]So all this is very new and technologies have not developed or not as much technologies the offerings that enable enterprises to easily use AI agents are not there yet but it will take time.

[00:11:41]>> So and when you talk about the agents too one interesting idea that you mentioned in your research note on the hype cycle is agent washing. what does that really look like in practice and how can leaders distinguish between real agentic AI capabilities that you just talked about and kind of the repackaged automation. So agent watching is this aspect where you know vendors used to sell solutions even before AI agents especially with automation or you know RPA.

[00:12:14]>> Now you can see that many of them have been rebranded as AI agents or agentic solutions.

[00:12:20]So the change is really in the marketing with a light touch of AI but they may not necessarily have much AI agentic uh capabilities that I mentioned earlier.

[00:12:32]Now it is important for you know you as you invest in AI agents to be really watching out for whether this is technology that's been rebranded like an old wine in new bottle or it's really allows you to build solutions that can make take actions or make decisions or go after goals that is something to clearly use to distinguish from prior RPA onto AI agents.

[00:13:01]What you want to do is you want to identify places to use AI agents that is beyond the scope of what an RPA can do but at the same time not so hard >> that a agents cannot in state of maturity today.

[00:13:18]>> I know you talked earlier too about the potential and there's real potential with these agents. Looking ahead, we know that every technology eventually moves into that trough of disillusionment.

[00:13:30]What do you think might trigger that moment for agentic AI and how should organizations be preparing for that?

[00:13:37]>> Yeah, it's a this is always happens with technologies as they go through hype cycle but they can take different times uh to come through the trough of disillusionment. The trough disillusionment is when the height starts coming down and because of which there's a lot of disappointments. So in agentic AI we may not be too far from it as people realize that while the potential exists it is not as easy to apply agentic AI >> and it requires underlying investments and process and data and when people realize that integration is a challenge security guardrails governance is a challenge and on top of it when people realize in some cases while we think agent AI can solve problems it really is not too much better than current technologies. All these factors can contribute onto it going through a trough of disillusionment. That doesn't necessarily mean the technology has does not have potential. What it just indicates is that we might have had too much expectations driven by hype in the media and in other places and then that has to come down. But after that as we get real use cases people become more comfortable with using AI agents as you have technology components that are ready for governance guard rails etc. Then you will start to see real value being delivered in businesses worldwide using this technology.

[00:15:07]>> For the second part of that, anything organizations can prep can do to prepare for that trust.

[00:15:11]>> Absolutely. While hype and agentic AI is high high, that does not mean that you should wait. The technology is mature enough for you to start thinking about where you can apply. This is what I tell clients. You want to identify places where agentic AI can deliver unique value. That might not be just assisting humans or automating work, but think about work that humans cannot do easily or quickly whereas an agent can, but it's still of business value to you.

[00:15:46]This might require us to stretch our imagination and work with our business partners, use internal challenges to identify some places where the work is more than what an RPA can do, but it's not too hard and do some of that work and through that you will learn and through that you might build frameworks that you can apply in multiple aspects of your business and then grow over time. This is the central point. The focus should be on identifying patterns that deliver value and doing more at scale and not as much in just trying to automate something and make it completely autonomous. That is not the sign of progress. The sign of progress is achieving value at scale across your organization even if we keep some human in the loop. And before we hit the trough, if you were advising a CIO or tech leader today, and I think you've probably answered components of this in your last response, but are there any other practical first steps they should take right now to get the value from Aentic AI without getting too caught up in that hype?

[00:16:57]First identify a few use cases working with your business partners where you can apply agent care for true business value. Ensure that you are capturing the learnings both on the technology side on the business side and move forward. The irony is that many organizations think that they want to start and use agentic AI in places where there are inefficiencies and process and data.

[00:17:24]But unfortunately, agent AI works better where your process is good and your data is clean, not the other way around. Be very critical about identifying those places before you start to use agent AI.

[00:17:39]And then once you start to build some then you can take them and scale in different parts of the organization.

[00:17:46]>> Great tip and definitely counterintuitive.

[00:17:49]Rajes, this has been great. One last question to end on. Is there a book you'd recommend to our listeners? Either something related to today's discussion or just a personal recommendation of a of a book that you really love.

[00:18:02]>> Absolutely. It's uh uh you know um I tend to read a lot and uh over the last few months I've been reading books that are about intelligence but not necessarily about artificial intelligence. So one book that I really found interesting is this by this author called Edong. It's called an immense world >> in which he speaks about animal senses.

[00:18:28]And it's quite an interesting book about um animal senses and how they are different. And what a a key point I got out of that is while there are a variety of senses that are used by animals, we as humans are limited to understand only those that our senses can make use of.

[00:18:48]So the world is filled with a rich type of intelligence around that some of which we ourselves may not be aware of because of our own limitations. I found that book quite interesting. I hope you enjoy it too.

[00:19:00]>> Rajes, thank you so much for joining us today. We really appreciate it.

[00:19:04]>> Thank you for having me, Karen. It was very good to speak with you.

[00:19:07]>> And thanks to all of you for listening to this episode of Thinkcast. To learn more about the 2026 hype cycle for Aentic AI, follow the links in the description. Thinkcast will be back where you listen to podcasts a week from today. In the meantime, be sure to rate, review, subscribe, and share with a colleague so neither of you will miss it.