By 2027, more than 40% of agentic AI projects are expected to be discontinued, according to tech research firm Gartner. The reasons? Poor risk planning, unclear business benefits, and rising implementation costs.
Despite the excitement around this new AI frontier, many early projects are struggling to scale or deliver value.
Many Projects Are Still in Experimental Stage
Anushree Verma, Analyst at Gartner, warns that a large share of agentic AI initiatives are still in their trial phases. Many are driven more by buzz than by business needs.
She notes that companies often overlook the hidden challenges, including the high cost and complexity of deploying agentic AI at scale.
Survey Shows Cautious Optimism Among Businesses
In a January 2025 survey of 3,412 business professionals, Gartner found:
19% had made significant investments in agentic AI
42% reported small-scale or cautious investments
8% had not invested at all
31% were undecided or waiting to assess the landscape
These numbers show that while interest is growing, uncertainty still looms large.
“Agent Washing” Clouding the Market
Gartner also flags a worrying trend: “agent washing.” This occurs when vendors rebrand older technologies—like chatbots or RPA tools—as agentic AI without true agentic features.
Only around 130 vendors, says Gartner, currently offer genuine agentic AI capabilities. The rest risk misleading customers and inflating expectations.
Real Agentic AI Still Has Limits
True agentic AI can perform complex tasks, make decisions, and learn from outcomes. However, most current models fall short of these goals.
Verma stresses that many so-called agentic tools do not yet provide strong business returns or justify their costs. In many cases, companies don’t need full-scale agentic AI to solve the problems at hand.
A Glimpse at the Future
Despite the setbacks, Gartner sees a bright future for agentic AI. By 2028:
33% of enterprise software may integrate agentic AI (up from less than 1% in 2024)
15% of daily business decisions could be made autonomously by AI (currently near 0%)
This shift shows long-term potential—but it will take time, focus, and smarter strategies.
Don’t Retrofit—Redesign for AI
Integrating agentic AI into old systems can be technically difficult and expensive. Verma recommends redesigning workflows from the ground up with AI in mind, rather than force-fitting it into outdated processes.
She emphasizes focusing on enterprise-wide productivity gains—not just boosting individual efficiency.
Making Agentic AI Work for Business
To unlock real value, companies should use agentic AI where it truly improves performance. Examples include:
Automating repetitive tasks
Supporting smart decision-making
Assisting with simple information retrieval
The goal should be to boost speed, quality, and cost-efficiency at scale.
Agentic AI is a major leap forward in automation and intelligence. But without a clear strategy and ROI-driven goals, many projects may fail before they deliver results.
The advice from experts is clear: avoid the hype, invest wisely, and design your systems for long-term success—not just short-term excitement.
