Lately, I’ve seen a specific pattern emerge as organizations make AI claims. “We’re AI-first.” “We’re AI-native.” “We’re agentic.” The language is confident, forward-looking, and nearly universal.
The results are generally not.
Last year, MIT found that billions of dollars in enterprise GenAI pilots yielded nothing measurable. Gartner predicts more than 40% of agentic AI projects will be canceled by the end of 2027, and a recent Gallup survey showed that just 13% of U.S. employees use AI daily. Even “frequent use,” defined as a few times a week or more, sits at only 28%.
So there’s a gap. Leaders describe AI as the most important shift since electricity. Their teams are still deciding whether to open the tools.
The headlines are calling this an adoption problem, but I think the real issue is credibility.
SWAP ABSTRACT ANSWERS FOR HELPFUL FIXES
I spend a lot of time with customers across the U.S., Europe, and Asia. In those conversations, nobody talks about LLM architecture or multimodal reasoning.
They ask: How can I see which projects are at risk before they become a crisis? How do I save my team from spending hours every week manually building status reports? How do we prioritize hundreds of incoming requests without adding headcount?
Our AI communications should be based on the answers to these questions: pragmatic, rooted in reality, and genuinely helpful.
Our own research confirms this. In a recent study, 52% of respondents said accuracy was the most important quality in an AI tool. Speed came next at 47%, followed by ease of use at 46%. People aren’t looking for a flashy digital assistant that impresses in a demo and disappears when the work gets complicated. They want something that understands and improves their workflow, whatever that might look like.
DELIVER PROOF ALONGSIDE PROMISES
Saying “you need to use AI” in 2026 is like saying “you need to use computers” in 1986. We need to start getting much more granular to gain trust. The best way to do this is use cases.
For example, our marketing team wanted to reclaim 10-15% of their time. That meant mapping specific friction points and matching each one with the right AI capability. They exceeded the target, and now we’re scaling the same approach across other departments.
That kind of result doesn’t require a team of management consultants to measure. Some of the most telling signals are small: shrinking meeting durations, compressed approval cycles, and faster deliveries. Digital agency Jellyfish, one of our clients, saved three to five hours per person, per week using AI. Legal firm Kalexius, another client, cut time spent in status meetings by half with AI use.
These are the metrics that survive budget reviews and create benchmarks for real growth.
EQUIP AI TOOLS WITH REAL-WORLD CONTEXT
Most AI tools don’t fail because the underlying technology is bad, but because they don’t know enough about the business they’re supposed to help. They give generic answers based on publicly available information, when you need specific details from a unique set of circumstances.
That’s where work platforms with semantically rich, permission-aware operational layers can help, providing AI features that draw on millions of data points to answer queries accurately and accelerate every point of your unique workflow. It’s AI, without the blindfold of context barriers.
This is the dividing line between AI that sticks and AI that’s just an expensive experiment. When AI understands your data, your team’s habits, and your organization’s priorities, it stops being just another piece of software and starts becoming an integral part of operations.
On a more human level, every genuinely helpful response grows trust in the technology, smoothing and speeding up adoption.
BUILD A FOUNDATION FOR LONG-TERM CHANGE
The organizations I see gaining the most AI momentum are the ones that identified a specific friction point, matched it with the right tool, and built from there, connecting all the dots along the way.
It might be time for every leader making bold AI claims to reframe what they have to say: Where is this technology working today and how is it actually helping the user? If the answer requires a caveat, a pilot disclaimer, or a reference to a future roadmap, the credibility gap is still open.
Closing it means fostering that very human emotion: trust.
Thomas Scott is CEO of Wrike.
