Follow ZDNET: Add us as a preferred
ZDNET’s key takeaways
- Professionals believe AI isn’t delivering value.
- Too many organizations suffer from tool blast.
- Focus on strong use cases to deliver uplift.
As many as 91% of employees say their organization is falling short of reaching the value that AI can deliver, according to the 2026 Future of Professionals Report from global content and technology company Thomson Reuters.
The research, based on a global survey of 1,800 professionals from multiple sectors, shows a widening gap between AI ambition and reality, which is becoming an increasingly significant issue, Kirsty Roth, chief operations officer at Thomson Reuters, told ZDNET.
Also: Why replacing staff with AI backfires – and 5 ways smart leaders generate real value instead
Eighteen months ago, employees were eager to experiment with AI, and their bosses were keen to support these explorations. Today, it’s another story. Professionals are burning through tokens as they use AI, with their bosses concerned about the upward trajectory of their IT bills.
With MIT research suggesting that 95% of AI projects fail to deliver value, it’s unsurprising that companies are getting cold feet.
AI is hyper-fragmented and hyper-fractured
“People are starting to work out [that] these technologies cost lots of money, and they don’t necessarily see the value from them yet,” said Roth.
“And so the conversation has turned into the classic change management one, which is, ‘OK, we’ve got all this tech and all these tools, but how are we really changing our processes and the way we operate to be more effective?'”
Also: ‘Specialists aren’t required’ anymore: How to stay valuable in an AI agent workplace today
New emerging technologies, from agentic AI models to deep research tools, will continue to enter the business. As Boomi CEO Steve Lucas told ZDNET recently, the general state of play in AI is hyper-fragmented and hyper-fractured.
“Professionals now recognize a bunch of terms — frontier models, private models, domain models, open-weight models, agentic frameworks, agentic harnesses, and agentic loops — that didn’t exist a few years ago, never mind a few months ago,” he said.
Roth, who considered her firm’s research and drew on her personal experiences, suggested businesses and their professionals who deliver value from AI focus on two areas: well-grounded explorations and solid production use cases.
Support well-grounded explorations
Professionals responding to the research were clear on what their AI tools must do: safeguard confidential data (96%), ground outputs in authoritative content (94%), and produce explainable and defensible reasoning (90%).
However, two in five professionals (41%) who use AI at work said they don’t have access to high-quality tools.
Also: Companies embracing AI the most are hiring more people – including entry-level
Even when an AI strategy exists, execution often lags, according to the research. Just over a third (35%) of professionals in firms with a named AI strategy say the approach isn’t visible in their day-to-day work.
Roth said one explanation is what she called “tool blast,” where organizations push a broad selection of AI services to staff without a clear business outcome in mind. “I’ve heard people say, ‘I’ve been given all these things. But what am I meant to be doing?’ Too many firms aren’t clear on what tools people should use. I think it’s then very hard to see the uplift, other than you’ll see your software costs go up significantly.”
Also: AI is getting better at your job, but you have time to adjust
As ZDNET reported recently, tech analyst Gartner predicted that 40% of enterprises will demote or decommission autonomous AI agents by 2027 due to concerns about value.
Roth said smart business leaders look for workable AI solutions to intractable challenges by giving professionals room to explore emerging technologies, without taking on too much risk.
That was something she did at Thomson Reuters, where the firm took what she described as an open-minded approach to generative AI.
“If you had a cool tool and [were] in marketing, sales, or coding teams, and you wanted to try it, we would let you try it,” she said.
Also: The new enterprise AI expert every company needs – and why
Roth said rather than go out with cost targets in mind, the business has encouraged people to test AI tools and see if they can find a better way of working.
“We’ve said to people, ‘Here are the tools. Reimagine what you can do,'” she said. “Now, obviously, some have done better than others, and some have needed more nudging and help than others. But we have encouraged people to think about what these tools can do in today’s world. And that approach seems to have worked well for us.”
A crucial element of this strategy, Roth suggested, is being clear about which tools do and do not provide benefits, and in the latter case, bringing explorations to a halt.
“Early on, we’d try anything for about six weeks. You could get a license to pretty much anything you wanted, and have your team play with it, depending on what kind of function you were in,” she said. “We’d test for the results, and if the results were good, we’d roll it out in other teams. And if the results weren’t good, we’d kill it, and we’d move on. So, I think that strategy of access and play early on is really important.”
Define solid production use cases
Roth said the companies pulling ahead in generative and agentic AI are those that turn explorations into production-level services, while the laggards do not.
“I think the successful firms are now starting to say, ‘OK, we have chosen, and we are going to use this tool, and therefore we’re all going to change our business process to operate in this new way,’ and then you start to see some improvements and savings,” she said, before adding an important caveat. “However, it still feels like the majority of firms are in the playground, if I can put it like that. I think those that are doing well are being very specific on their use cases.”
Also: How Workday and other software providers plan to survive AI
At Thomson Reuters, the specific use cases focus on five key areas: engineering, customer support and success, marketing, editorial and content operations, and core technology operations.
“Of course, we’d love to make everyone’s job better, and we will do our best,” she said. “But these are the five big areas where we see the biggest uplift, and they get the most focus.”
Across these areas, AI tools are found, tested, and deployed, and then their effectiveness is measured.
Today, 87% of Thomson Reuters employees actively use AI tools in their daily work. So, what does successful generative and agentic AI look like across the organization, and how has it changed employees’ day-to-day practices?
Also: AI agents are your new colleagues – how to get the best results
Roth gives the example of customer and sales support, where staff can use the firm’s internal AI platform, known as Open Arena, and the frontier model Claude.
Rather than spending hours collating information from sales reps and the Salesforce platform, staff can use approved AI services to receive answers to queries in seconds.
“In the new world, you can write a prompt in Claude to pull that information; it can write you a summary, understand what the key opportunities are, where there might be any risks to that account, or whether they’ve called support recently, and they were irritated about something, and you’re well prepared for that meeting far more quickly,” she said.
Thomson Reuters employees also use AI for market research, document writing, and tracking the profitability of products and services.
Also: The 3 types of people who will excel in the AI agent era, according to tech leaders
Having spent the past few years embedding AI into the operational reality of a 27,000-person global workforce, Roth said the key thing she’s learned about putting AI into production is that business leaders must work hard to overcome professionals’ fears.
“Humans don’t like change. Demystifying AI early on was key, and then it was just about letting people have a chance to play with things and hopefully not be afraid of them in the way we teed the services up,” she said. “Now we’re getting to a much more mature state and, with what we’ve managed to adopt, I think the direction of travel and the consistency are equally important.”
Artificial <a href="https://myappsplus.com/intelligence-leaders-eye-automation-for-faster-targeting/” title=”Intelligence Leaders Eye Automation for Faster Targeting”>Intelligence
- How to avoid Claude watermarking your content
- Taught by AI pioneers, Stanford’s free online course takes you far beyond ChatGPT
- 75% of developers I surveyed prefer Claude Code – here’s why they choose it over Codex
- How to keep your conversations with ChatGPT, Gemini, Copilot or Claude as private as possible
- My 3 favorite AI tools for voice dictation while vibe coding – and one is free
