Senior executives from five technology companies outlined the next stage of artificial intelligence in business use. Their comments point to a shift from experimentation to trust, integration, and measurable results.
Common themes emerge throughout the responses. That means companies are moving beyond their initial enthusiasm for generative AI tools and starting to ask whether they can trust the technology for their core operations. Several executives said the debate is no longer about simple adoption, but whether AI can reduce workloads without creating new risks or adding checks.
Dharam Gurbani, chief growth officer at Ascendion, said organizations are now testing AI based on actual business demands rather than broad strategic arguments.
“In the UK and the rest of the world, the real test of AI is not ambition, pilot or strategy; it is whether AI will make critical work more reliable in the organizations that people rely on every day. Across the UK businesses we work with, leaders have stopped asking if they can deploy AI. They are asking harder questions. How can we improve compliance, reduce operational burden, strengthen compliance, and improve customer experience? The answer is that AI needs to prove itself within a company’s operational structure, not just be fixed.” In practice, that means modernizing outdated systems, eliminating bottlenecks, facilitating compliance evidence, and delivering more high-value work to teams. This means empowering teams to move faster while maintaining strong governance and judgment. In other words, AI is worth evaluating. ”
The emphasis on trust is also evident in Foxit’s comments, which highlight how much time workers still spend checking the output generated by AI. The study found that while the majority of executives see productivity gains from AI, those gains can be undermined by validation efforts.
“Of course, AI Appreciation Day is an opportunity to recognize how quickly AI is evolving. But let’s also take the time to consider whether organizations actually trust the jobs AI is producing. Our latest research shows that 89% of executives believe that AI is improving productivity, but that improvement is often undermined by the time spent validating the output produced by AI. In fact, executives use AI to increase productivity by an average of 4.6% per week. You’re saving time, but you’re spending about the same amount of time checking that work. If you don’t have confidence in the output, you’re less productive.” In recent years, companies have had to move beyond AI adoption to trust in AI. The bigger challenge will not be determined by how many AI tools companies deploy, but by whether AI agents become more capable and able to tackle more complex tasks. Trust becomes a decisive competitive advantage, starting with the quality of information provided by AI and supporting governance and verification processes. Organizations with these foundations in place will be able to enable employees to spend less time checking AI and more time building on its insights. ” said Evan Reiss, Senior Vice President of Marketing and Innovation at Foxit.
core process
Derek Thompson from Workato argued that many adoptions are still limited to basic tasks such as summarizing and editing, leaving broader business value unrealized. He said the larger impact will depend on whether companies can allow AI systems to operate within their core business processes and measure them against defined metrics.
“Despite the rapid growth in AI adoption, the majority of use cases to date have been fringe experiments, such as simple tasks such as summarizing research or rewriting emails. By limiting technology to surface-level tasks, you are also limiting the impact technology has on the surface-level business. For deeper integration and deeper impact, CIOs need to trust AI in core business processes. By aligning the use of AI with the metrics that matter most to the business, CIOs can empower every agent to drive results with measurable impact. The success of agent AI is determined not by the sophistication of individual agents, but by how effectively organizations can integrate them into their business structures. ”
OutSystems took a similar view, but focused on narrower use cases where AI can already support staff in routine and data-intensive tasks. Lewis Brand said companies see more value in systems that support human decision-making than in fully autonomous tools.
“Despite the hype around fully autonomous systems, most businesses today are using AI in far more practical ways, and that’s where the real value comes in. The most powerful use cases are centered around three areas: processing documents that require human review, processing high-volume transactional tasks such as mapping incoming orders, and supporting decision-making by making sense of unstructured data. In these scenarios, AI It’s great at summarizing complexity and making recommendations, but it’s not great at making final decisions unless the organization is willing to buy into it. Used poorly, AI can act like a team of interns. Still, it requires monitoring and double-checking. When used well, especially when leveraged with the right data and guardrails, trust comes from knowing when to help and how the two work together.”
Model selection
Vercel’s Malte Ubl mentioned another issue shaping enterprise use of AI: cost. He said the team has become more selective in its model selection, using lower-cost systems for high-volume work and reserving more advanced models for tasks where accuracy and reliability are most important.
“Days like today typically bring one of two conversations to the surface: how fast the models are improving and what could go wrong. What we can see in the data is more realistic. Teams are starting to be more careful about where they spend their AI budgets and where they don’t. Even as costs rise, Vercel AI Traffic through gateways continues to grow. But the way teams spend money is changing. Cheaper models handle the bulk of the work, and the most capable models are reserved for tasks where quality, reliability, and accuracy really matter. “No model wins every job, so the team that routes each task to the right one is the one that understands which model is right for which task and balances quality, speed, and cost accordingly.” said Malte Ubl, Vercel’s chief technology officer.
Taken together, these comments suggest that a more restrained phase of AI adoption in enterprises is taking shape. Rather than asking how broadly AI can be deployed, companies are increasingly asking how much AI can be trusted, how it should be monitored, and whether the results are worth the cost and effort.
