Artificial Intelligence has officially reached the “you can do anything with it” stage of technological hype. Judging just by AI vendor narratives, and by some of the media, it’s the golden key to productivity, efficiency, optimization and innovation. And if you add agents? They will think, write, negotiate, design, respond, validate automate, and probably brew your coffee and pick your tie, if you wear one. All you need to do is… well, everything else. And clients are trying, with 59% of organizations in Europe declaring they are using Agentic AI.
Because while vendors keep promising moving mountains with a few lines of AI code, clients are still trying to figure out how to move their own data out of legacy systems and into AI.
The Vendor: Here’s your AI, you can do anything with it!
Two years into generative AI world, and still, you walk into any tech conference, and you can hear the same verse of a song we all know: “With (our) AI, you can/must/need/should/want to transform your business!”.
And it does sound wonderful until you realize that behind the shiny demos and glossy ppt decks, there’s a subtle assumption that it is the client, who will have to make it happen.
The vendor provides the vision; the client does the homework.
And let’s not be fooled, it is not just a little homework. We’re talking about a full school year project, that, if you’ve ever done it, you’re dreading; from data preparation, to process standardization, to employee training, to infrastructure modernization. Let’s not forget we also need to build convincing business cases for the board. The same board that may also add to the noise with inflated expectations, like parents tricked by the school with a promise of great grades and a ticket to the best university, if only your child works hard enough.
The Client: We’d love to, if only we had time…
When you talk to most enterprises, you’ll probably hear a slightly different tone than that of vendors. Clients are not rejecting AI; they are overwhelmed and exhausted by it: 80% of clients in Europe say they manage to move beyond the PoC stage, but only half of those projects bring measurable outcomes. They’re juggling existing priorities often squeezed by budgets (even if AI spending seems to be relatively immune to budget cost, as we see in IDC research). And they are asked to lead AI transformation projects as if they had a spare team of data scientists and engineers hiding somewhere.
And most painfully, vendors are often seen as teachers not that willing to help. As one manufacturing executive put it: “They come excited, but when asked for business cases or scenarios we can use, there’s silence”.
That silence speaks volumes. Clients know that AI isn’t plug-and-play, it is much more complex and time-consuming. You need to clean, prep, build, deploy, test and hope it scales. They understand that before you even dream of “autonomous decision making,” (though the sheer promise of autonomy can also scare some clients!) you need data consolidation, process standardization or cross-department alignment. Without that backbone connecting all stakeholders, no model will do magic. And oh, yes, change management, anyone? This needs to be planned and taken care of, too. Plus, once you have done all the work, you realize, this work continues: 49% of companies today focus also on making existing AI projects work better.
But that’s rarely what the vendor pitch slides say.
Only we are not at school anymore…
Let’s agree, for many clients this feels like school all over again. Vendors show up as enthusiastic teachers, armed with the latest learning materials (“Look, a new LLM!” or “How about a fresh set of agents?”), and clients are expected to do the homework: research, structure, and present working examples at the next review.
The difference? In school, a good teacher stays after class to help you understand the assignment. In AI world, too many vendors drop the textbook on a client’s desk, or a presentation into their mailbox, and head to the next classroom, I mean, to the next client…
What is really needed is a different educational approach. Good teachers, like good business partners, guide, not grade. They adapt to the student’s (or client’s) level, pace, and context. They explain the “why” before demanding the “how”. That’s what’s so painfully missing in much of today’s vendor-client dynamic: tutorship instead of lectures.
The missing link between a vision presented and the value expected
The truth is, you cannot “do AI” without doing groundwork, in that sense, homework is needed. You are as good at math as the number of math problems you solved. And sure enough, that groundwork is not glamorous.
And this much needed AI readiness isn’t about another platform or toolset; it’s about structure, sense, and support. The real hard work happens behind the scenes: making data talk to each other, mapping or streamlining processes, and, most importantly, preparing people to trust and use new systems. It’s less about technology as such, and much more about maturity: operational, organizational, and yes, managerial.
To achieve that, what companies need most are partners who understand their business language as well as their data models. As one CIO put it: “We need partners, but we must drive the strategy, not follow vendors”.
And that’s precisely where the conversation must change. We, or the market, desperately need to shift from unrealistic ambition – AI for all! to very realistic accountability – AI that fits.
Going forward less pitching, more partnering, please…
For vendors, this gap should be more than a talking point – it’s a wake-up call. The distance between what’s promised and what’s practical keeps growing, and clients are losing patience and hope. We’ve all heard the same speeches about collaboration, co-innovation and customer-centricity and yet too often, the market still gets more promise than substance.
Clients aren’t lazy students skipping their homework. They’re professionals keeping production lines running, supply chains stable, and customers satisfied and all while being told they’re not “AI-ready” enough.
If vendors want to stay relevant, they need to move from selling solutions to solving problems. That means stepping off the stage, where they are fighting with competitors and into the trenches to fight for clients, learning the nuances of each industry and individual organization’s needs. It also means realizing that success isn’t measured in the number of features added, dashboards set up or models deployed, but in business outcomes that actually matter.
This may sound like a cliché but still true: at the end of the day, clients don’t buy AI (or any other technology for that matter); they buy results.
We all need less hype and more help
The promise of (well executed) AI is real. But so is the fatigue with AI at this point. It’s time to close the gap between “you can do anything” and ” but you need to do everything yourself”.
If vendors truly want clients to embrace AI, they must act less like teachers assigning homework – think Bismarck-style education, and more like tutors walking beside their students – think Socrates before he drank a hemlock infusion. It’s this moment, when we realize, unlike at school, there’s no final exam in business, only continuous learning.
And I would bet those who help their clients learn with them, not for them, will be the ones still standing when the hype cycle fades.
You will hear more about AI and agents in IDC EMEA’s FutureScape Predictions webcast this December. You can register here,
You may also be interested in a webcast Ewa is presenting on AI Sovereignty. You can register for that one here.
For an overall look at AI in EMEA, you can download this eBook.
To learn more about how International Data Corporation (IDC) can support your technology market data needs, please contact us.
