Gartner says add AI agents ASAP – or else. Oh, and they're also overhyped

Follow ZDNET: Add us as a preferred source on Google.

ZDNET’s key takeaways

• Gartner predicts 40% of apps will add AI agents by 2026.
• Business leaders face hype-driven pressure to act within months.
• AI value is real, but rushing adoption is dangerous.

Pssst. Hey. You. Yeah, I’m talking to you. Are you a CEO, board member, senior VP, or other top-level corporate leader? You want to know a secret?

You’ve got three to six months to AI agent-up your company, or you’ll fall behind. You know what that means, doncha? If you fall behind, you’re out. 

Also: 95% of business applications of AI have failed. Here’s why

This is the gist of a highly questionable forecast coming out of Gartner this week. As part of the analyst firm’s predictions on agent adoption in enterprise apps, the researcher claims this: “CIOs have a crucial three- to six-month window to define their agentic AI strategy, as the industry is at an inflection point. Organizations that do not embrace agentic AI promptly risk falling significantly behind their peers.”

What does that even mean? Falling behind how? The key selling pitch is that agents can do more and cost less. So, is the big theme here that if you don’t dump a pile of employees and replace them with AIs, you’ll spend more than your peer companies? Or is there some expectation of innovation in three to six months?

Let’s deconstruct this, and then add some more details from Gartner’s report.

The promise and peril of AI agents

First, there’s no doubt that autonomous AI agents have some potential for increasing productivity and value in business. But they are shaky as heck right now. For example, I used ChatGPT’s premium $200/mo Pro account to test OpenAI’s brand new Agent mode. Out of 8 tests, only one returned any value at all.

Also: Gen AI disillusionment looms, according to Gartner’s 2025 Hype Cycle report

I ran a few more tests and did manage to find some more value. In one of the additional tests, I used Agent combined with NotebookLM to do some research, and the result was very helpful. I also used GPT-5’s Deep Research in Pro mode to do some code analysis, and that was helpful as well.

But we’ve also seen that agent coding in GPT-5 is fairly terrible, resulting in both hallucinations and what the AI itself admitted were “unconscious” assumptions.

Stages of agentic AI evolution

When Gartner doesn’t get caught up in press-pandering hyperbole, it makes some potentially valid points. For example, it identified stages of agentic AI evolution for the next five years.

1. 2025 – AI assistants for every application: Adding AI through an LLM API is an easy coding challenge, fairly inexpensive to implement, and provides a new profit center. So sure. Every app vendor who can figure out a pitch for adding AI to their app will, whether it needs it or not.

2. 2026 – Task-specific agent applications: Enterprise apps will start to add task-specific agents who can handle narrow responsibilities. This is a reasonable assumption, as long as the AIs behave themselves, and the tasks are specified clearly and completely.

3. 2027 – Collaborative AI agents within an application: This is the idea of building teams of agents that work together to perform complex tasks within enterprise applications. This is also reasonable for certain specific types of tasks and applications. The potential for major cascading failure is here, too.

4. 2028 – AI agent ecosystems across applications: Agents within applications will talk to other applications. To some degree, this is an extension of the API or microservices idea we’ve had for years, but with some smarts added.

5. 2029 – “New normal” of enterprise applications: Gartner says, “Agents will be created on the fly by humans, and humans and Al will collaborate in new ways.”

Let’s make a prediction, shall we? Agents created on the fly (which implies a lack of thought and lack of planning) will result in some very bad outcomes. This is not a goal. This should be a cautionary tale. When Gartner says that 50% of “knowledge workers” will be able to work with the AIs and create agents, that’s plausible. But on-the-fly rapid deployment? That’s how you get Skynet.

Gartner’s headline prediction is that 40% of enterprise applications “will feature task-specific AI agents by 2026, up from less than 5% in 2025.” The analyst firm also predicts that agentic AI will “drive approximately 30% of enterprise application software revenue by 2035, surpassing $450 billion, up from 2% in 2025.”

Gartner’s mixed messages 

Okay, fine. But earlier in the month, Gartner said that AI agents are at the Peak of Inflated Expectations and headed for the Trough of Disillusionment next. We also know that 95% of business applications that have tried to use AI have failed.

Also: My 8 ChatGPT Agent tests produced only 1 near-perfect result – and a lot of alternative facts

These are conflicting numbers and conflicting messages. That’s because hype and reality don’t always align. What makes things worse is that when there are glimmers of reality in the hype, the hype becomes all the more believable. AI is that way. Yes, there’s a lot of hype. But there’s also an amazing amount of value and innovation. But there’s still tremendous hype.

My beef with Gartner isn’t its forecast. It’s the pressure some of its statements put on decision makers. For better or worse, corporate leaders take what Gartner says as business guidance. When that guidance is predictive, it’s quite helpful.

But when that guidance incites an alarming sense of urgency, as “falling significantly behind their peers” does, it pushes all the wrong buttons. Business leaders never want to fall significantly behind their peers. That implies reduced earnings at best, and landing on the unemployment line at worst.

Statements that provoke business leaders to push through initiatives in three to six months, most likely without the proper level of deliberation, caution, and impact analysis can cause serious harm. 

Also: 8 ways to write better ChatGPT prompts – and get the results you want faster

So, what’s a business leader to do with these mixed messages? Do your due diligence and don’t let the hype machine pressure you into risky, rash decisions, for starters.  

Do Gartner’s predictions reflect a realistic timeline for agent adoption, or do they place too much pressure on leaders to act quickly? How can companies embrace the benefits of AI while avoiding rushed decisions? Let us know in the comments below.

You can follow my day-to-day project updates on social media. Be sure to subscribe to my weekly update newsletter, and follow me on Twitter/X at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, on Bluesky at @DavidGewirtz.com, and on YouTube at YouTube.com/DavidGewirtzTV.