The ambition to execution gap is widening

As strategy accelerates, modernization is the breaking point between momentum and stagnation. It’s widely seen as a “safe” investment—leaders know they need to modernize—but as Berg notes, many efforts “focus on migration, not value.” Too often, data is moved into modern platforms without asking which data will truly drive differentiation.

Nearly 8 in 10 companies (77%) are investing in modernization, but the depth of those efforts varies. Only 34% have an enterprise-wide mandate to fully replatform, and 42% admit their roadmaps are only partial—efforts that move the needle but don’t go far enough to transform how the business works. More than half of organizations still run most core applications on legacy systems, and even among those with active modernization programs, only 42% say they have the robust data foundations required for meaningful impact. Together, these signals show how incomplete modernization is one of the main forces quietly limiting AI’s potential. 

These gaps reveal a fundamental challenge: true modernization isn’t just technical—it’s about making data usable, connected, and aligned to business purpose. That level of change demands both clarity of direction and disciplined execution, and many companies still struggle to balance the two.

“Everything we dream of doing with AI is grounded in data,” says Amalia Goodwin, managing director of adaptive organizations at Slalom. “You can’t make confident decisions or create real impact if your data isn’t trusted, accessible, and tied to your strategy.”

AI’s potential depends as much on trust as on technology. Modernization may give companies a sturdier data foundation—but without security, transparency, and ethical oversight, that foundation is still vulnerable to risk.  

As AI becomes increasingly embedded in critical workflows—95% of leaders say they’re comfortable relying on copilots for strategic work—governance and trust aren’t keeping pace. Only half of companies have formal audit or ethics boards to oversee AI. Most focus on technical safeguards, like real-time monitoring or AI-tuned cyber-threat detection (45% ), while fewer (42%) have built model and data controls that include bias, drift, or lineage monitoring. 

The rest rely on fragmented oversight: 29% at the business-unit level, 9% with vendors, and nearly 1 in 10 still have no AI risk controls at all. Without clear accountability, responsibility falls through the cracks—and when failures happen, it’s people, not systems, who pay the price.

Even where controls exist, they focus narrowly on compliance rather than ethical foresight or stakeholder engagement. Only 38% of respondents say they engage external stakeholders in risk governance. This limited oversight reflects a familiar growing pain of early maturity: some are still building the capacity for responsible oversight while others assume existing controls are enough.

Usher describes this phenomenon as a kind of “human hallucination”—an overconfidence in people’s ability to self-police risk without the necessary systems in place. Many assume responsibility ends with compliance, she notes, when it should extend to continuous, transparent oversight. 

“You can’t delegate trust to the system,” says Goodwin. “Leaders have to own it—across data, decisions, and the people who use them.”

Without stronger safeguards and shared accountability, companies gamble with safety as they scale AI adoption. Building securely has never been just about compliance; it’s about giving employees and customers confidence that the systems shaping our world are safe because they’re built to be. 

If security builds confidence in AI, experience builds connection. Both depend on the same foundation: systems and processes designed for people, not just performance. Layering new strategies onto old workflows may deliver short-term efficiency, but it rarely unlocks new forms of value.

More than half of companies (56%) are redesigning workflows to automate or hand off tasks to AI with tangible gains: 62% report less manual work, 41% faster decisions, and 38% higher-quality outputs. Yet impact remains limited: just 21% of use cases scale enterprise-wide, and only 39% have achieved orchestrated multi-agent workflows across end-to-end processes. In most cases, automation is still applied to isolated tasks rather than full value chains—producing quick wins in productivity but little transformation in decision-making, collaboration, or customer experience.

Ownership is often the missing link. Cross-functional leadership for AI has expanded sharply—63% report shared ownership, up from just 5% in 2023—but accountability remains diffuse. Thirty-eight percent delegate risk decisions to business units and 18% outsource them entirely. When responsibility for integrating AI into daily workflows is unclear, process redesign stalls, and experimentation rarely matures into execution.

“To truly transform, you have to stop automating yesterday’s work,” says Goodwin. “Redesigning how decisions get made, where humans add value, and how teams interact with AI—that’s where scale happens.”

The bottom line: Real transformation will require radically reimagining processes, not retrofitting AI into existing ones.

AI is reshaping the social architecture of work. Many leaders feel prepared, but few are investing effectively in the skills, coaching, and organizational models needed to sustain change.

Two-thirds of leaders and employees (68%) believe their organization is keeping pace with AI, yet nearly all (93%) report workforce barriers—from skill gaps and limited time to inadequate training and tools—that limit progress. Over half (55%) identify workforce skills as the top obstacle to realizing AI’s value. While 69% say AI has played some role in reskilling efforts, only one in five describe those initiatives as very effective. 

And even when programs exist, few scale through sustained support: just 39% of mid-level leaders feel equipped to coach or mentor their teams, and only 9% of employees report receiving such coaching. The result is training without traction: programs that raise awareness but rarely build lasting capability. 

As AI reshapes work, the definition of “critical skills” is shifting too. Half of leaders now prioritize critical thinking and problem-solving, while fewer cite communication (22%) or empathy (19%) as essential—signaling a pivot from interpersonal to analytical capabilities.

The workforce is already being reconfigured around this shift. Forty-five percent of companies are creating new AI-related roles, and 36% are consolidating existing ones. Yet only about one in four (23%) report that net workforce reductions are already happening or expected within the next 12 months—evidence of reorganization, not replacement. As roles evolve, companies must move just as quickly to redefine how people learn, adapt, and stay effective in the flow of change.

“The companies that build learning into how they work—not just what they teach—will be the ones that adapt the fastest,” says Usher.

Strategy, data, security, process, and people are essential—but business value is the outcome they’re all designed to achieve. And expectations are high. Ninety percent of companies plan to increase AI investment in 2026, up from 81% in 2025 and 70% in 2024—a 20-point rise in just two years. Only 1% expect to invest less, suggesting that AI budgets remain resilient even as scrutiny on returns grows. When engaging external partners, value still drives decisions: 70% prioritize cost efficiency and clear ROI, followed by speed to value (63%), access to advanced technology (49%), and risk mitigation (46%).

Yet few are equipped to prove that value. Only 39% have clear metrics and  ROI tracking for AI performance, even as 68% expect returns within two years. Compressed timelines and limited measurement create pressure for quick wins and make longer-term impact harder to prove.

Most leaders still frame AI in terms of efficiency (76%), while fewer see it as a catalyst for reinvention (48%) or evolving customer strategy (30%). Only 38% say AI is delivering higher-quality, more consistent outputs—a sign that productivity gains have yet to translate into distinctive performance or tangible customer impact. By focusing on internal efficiency, many companies risk missing where disruption is actually forming—among emerging players building AI-first strategies from the ground up.

“Everyoneʼs looking at the same incumbents when the real disruption is still in stealth mode,” says Goodwin. “We will likely start to see the next wave of AI-native competitors emerge in 2026-2027.”

The takeaway? AI’s short-term value story is about productivity; its long-term potential lies in reinvention. The leaders who measure both will be the ones who convert efficiency into enduring differentiation and customer relevance.