To explore what’s really happening inside leading companies, we surveyed the Catalant Practice Community, our curated group of elite independent experts who operate where strategy meets implementation. These on-the-ground operators see the true “reality-to-hype” ratio. They are the ones currently advising executives on where to place their bets to ensure a significant return, partnering with functional leaders to ensure rollout success, and cleaning up the discarded pilots that failed to scale. This report distills their collective intelligence to explore how companies are actually deploying AI, what’s going well, and where they are falling short.
Setting the scene
The Current AI Landscape
Today’s environment is defined by urgency and uncertainty. Leaders are under immense pressure from boards and investors to “do something with AI,” yet teams are often paralyzed by the velocity of evolution and overwhelmed by the dizzying array of new AI products going to market. To examine this challenging landscape, we looked into how organizations are prioritizing their AI investments and the specific problems they are desperate to solve.
AI priorities and value drivers
Current AI investment is strikingly narrow. When we asked consultants what primary value drivers are motivating AI initiatives within their client organizations, the response was a landslide: increasing individual employee productivity/efficiency.
Value Drivers Motivating Client AI Initiatives
This concentration makes sense: productivity is seen as a tangible benefit and it is the natural result of plugging AI into existing workflows and processes without re-engineering the entire business. Especially for companies that are earlier on the AI maturity curve — those that are exploring and experimenting, trying to demonstrate quick wins, and looking to gain ground quickly — prioritizing productivity is a relatively straightforward place to start.
However, a notable second priority emerged: keeping up with competitors. This reveals the psychological weight of the current hype. Many leaders feel an existential pressure to deploy AI, even if the use case is fuzzy, simply to maintain competitive optics.
Key takeaway
Many companies are currently prioritizing “defensive AI” over “offensive AI,” but this is likely to be a short-term approach. When companies are only using AI to do the same things faster, they are gaining temporary parity. Market leaders are already shifting toward offensive AI strategies — using the technology to reinvent business models entirely.
Capabilities and leadership
To get a closer look at the types of applications companies are focused on, we asked what capabilities are being prioritized by clients. While the appetite is broad and encompasses all types of AI solutions, two areas stood out: analytical AI (data-driven insights and forecasting) and generative AI (content, code, and collateral creation).
Prioritized AI Capabilities
These results highlight the natural entry points for AI where many companies are still gaining their footing. Analytical tools democratize access to data and insight, empowering contributors at all levels to make faster, data-driven decisions. Meanwhile, generative tools offer immediate impact with low friction, enabling individuals to do more with less.
In addition, our consultants indicated that functional leaders are currently most often the people driving the AI ship, with CEOs and other leaders trailing by a significant margin.
Primary Driver of AI Strategy
Especially as a large portion of AI initiatives remain focused on productivity gains, it is logical that functional heads have been at the helm. Effectively pivoting to an offensive posture will demand tight cross-functional orchestration. Functional leaders often lack the authority to break down cross-departmental data silos or mandate enterprise-wide changes.
Without a cross-functional leader like the CEO taking the wheel to align vision and strategy, organizations end up with isolated pockets of innovation that fail to move the needle.
the path to success
Strategy and Value
If the landscape is cluttered with “AI theater,” what does the path to actual success look like? We asked our community to identify the foundations of effective implementation and how they have seen clients successfully measure results and prove value. All of this starts with looking at AI strategy.
Strategy as the foundation
The verdict from the field is clear: 84.4% of consultants agree or strongly agree that AI strategy must be driven by business pain points, not feature sets. The best AI strategy is business strategy amplified by AI.
Organizations often fall into the “shiny object” trap, where leaders see a new AI tool or capability and go searching for a problem to solve with it. Strategic implementation is the inverse: it starts with a vision of how the business should innovate or a specific bottleneck that needs clearing.
84.4%
of consultants agree or strongly agree that AI strategy must be driven by business pain points, not feature sets.
Execution is a human equation
Once the strategy is set, companies must be able to execute. To investigate how companies are effectively implementing AI, we asked consultants to identify the most critical factors for successfully scaling AI initiatives.
The answer was definitive: Success isn’t only about the technology—the most critical factor for successfully scaling AI is human change management and enablement. In fact, our experts overwhelmingly agreed: Success in AI is 70% people and process transformation, and 30% technology.
“Stop leading with technology. The organizations that win with AI in 2026 aren’t the ones with the best agentic harness; they’re the ones that rethought how work gets moved, brought their people along for the journey, and kept governance front and center. Tech is the easy part. People, process, and governance always were, and still are, the hard part.”
Beyond the top spot, we saw a three-way tie for second place:
Cross-functional teams with diverse skillsets
Engaged, attentive humans-in-the-loop for oversight
Iterative execution and tight test-and-learn loops
Most Critical Factors for Scaling AI
All of these top success factors have to do with humans: their buy-in, their collaboration, and their oversight. To ensure improvements are consistent and don’t negatively impact other business areas, you need a culture that embraces constant feedback and rigorous human oversight.
When it comes to AI, success is largely human-centric. Humans must learn and adopt the technology, ensure the quality of its outputs, and continuously strive for improvement. It’s important for leaders to zoom out from the technology and think holistically about AI initiatives.
Costs and value measurement
One of the most controversial questions we asked was whether AI should be treated as a non-negotiable cost — simple operational overhead — rather than a traditional ROI-driven capital project.
“AI should be treated as a non-negotiable cost of survival, rather than a traditional ROI-driven capital project.”
There is a baseline of AI investment that can now be considered the cost of doing business — driven by pressure from boards and investors to use AI and the reality that individuals are already using LLMs to work faster, so companies must provide safe and secure environments. There is also a massive potential opportunity cost to non-adoption that traditional ROI models fail to capture.
However, there is a dangerous trap here. Treating AI programs as non-negotiable without a clear value correlation is a recipe for runaway costs with no return. This is particularly true for specialty tools or AI systems where costs are flexible based on compute, data sources, or tokens used. Companies need a method of tracking spend and linking it back to defined business metrics and objectives in order to rationalize investments.
The question of how your organization treats AI spend needs to be explicitly discussed. Setting up frameworks for tracking, measuring, and optimizing AI costs from the start puts companies in a better position to make informed decisions and adapt when projects fail to deliver ROI.
When we asked how companies are successfully measuring value, the results were telling:
Time saved (process velocity)
Error and rework reduction
Hard dollars saved (OpEx)
Successful Methods for Measuring AI Value
The heavy focus on time saved aligns with the focus our consultants have seen on productivity and efficiency, where process velocity is one of the most clearly linked metrics. However, saving time isn’t always tied to boosting the bottom line.
The real question for leaders is: What are your employees doing with their “extra” time? If increased velocity doesn’t translate into higher-value output or reduced headcount, this improvement isn’t tied to real ROI.
Barriers and challenges
Even with the right strategy, the path to enterprise-wide AI rollouts and tangible returns is fraught with friction.
Our Practice Community consultants identified a wide range of early warning signs that AI initiatives will fail to reach their goals, including:
- A disconnect between leadership ambition and workforce reality
- Treatment of AI as a technology project rather than an operational transformation
- Unrealistic expectations for ease of implementation or “magical” transformation
- A lack of clear production KPIs and understanding of how to measure them
The greatest challenge facing leaders as AI gains traction is bridging the talent and skills gap. With the recent AI boom, even professionals who have worked hard to build AI fluency may not yet have true AI expertise. This creates a high level of competition for unicorn talent — experts who possess both the technical understanding of AI and the deep functional acumen to translate data and business needs.
But finding the right talent to lead the AI charge or serve as AI change champions isn’t the only hurdle.
Greatest Challenges for Leaders
Our consultants also identified significant concerns about aligning AI with business vision and the difficulty of scaling beyond isolated pilots. To understand why so many initiatives fail to make it out of the pilot stage, we dug deeper, asking specifically about the struggle of scaling pilots into production.
While consultants again flagged human issues such as workforce resistance and change management failures as significant challenges, this issue of scale is where data sat top of mind.
Most Significant Barriers to Scaling AI Pilots
Pilots thrive in clean sandbox environments. The move to production is where the reality of legacy systems and fragmented data sets emerges, making poor data quality, silos, and fragmentation a debilitating problem. AI has made data both more valuable and vastly more complex, and the classic “garbage in, garbage out” rule is magnified here; without a cohesive data strategy and strong data foundation, even the most sophisticated models will stumble.
Key takeaway
Moving from pilot to production is where companies are likely to discover they weren’t yet ready for the AI initiative they started — the data wasn’t clean, the governance wasn’t in place, or the team wasn’t bought in. This foundation needs to be established from the beginning to ensure initiatives maintain momentum and have the capacity to reach their goals.
Consultant Advice
Guidance from AI Leaders
To conclude our survey, we asked our community of consultants to distill their frontline experiences into the singular pieces of advice they believe are most critical for leaders to hear.
What emerged is a clear mandate: the technical hurdle is rarely the tallest one. Instead, the success of an AI initiative is dictated by the human architecture surrounding it. The following advice highlights a fundamental truth: you can have the most sophisticated models in the world, but without a transformation strategy rooted in transparent communication, deliberate buy-in, and the cultivation of momentum, your initiative will stall before it ever reaches scale.
looking ahead
What’s Next for Leaders
It’s important to recognize that the landscape is shifting again. We are moving from a world of “AI as a tool” (chat) to “AI as a teammate” (agents). In the near future, a leader’s span of control won’t just be measured by human headcount but by the orchestration of autonomous workflows. These agents will communicate with each other to execute complex tasks with minimal oversight.
Regardless of how fast this future arrives, the fundamental lesson remains: the real winners will be those who move from “defensive AI” to “offensive AI,” lead with a human-centric change strategy, and resolve their data debt.
AI is more than a software upgrade. It is a total redesign of how work gets done. This requires not just data and tools but also a focus on culture and training.
Move beyond AI theater to enterprise-wide advantage.
Connect with Catalant to engage with experts who have the AI skills and operational expertise needed to translate AI goals into tangible results.
This report is based on a survey of Catalant’s Practice Community, a global group of experienced and established consultants with an average of more than 20 years of experience as consultants and operators.
Survey invitations were sent to 751 consultants. The survey was open for 14 days and yielded 64 voluntary responses. Respondents identified their core functional focus areas and the areas in which they most frequently work.
All results reflect aggregated responses and are intended to capture directional insights into current AI initiatives and value drivers, successful AI strategies, barriers to scale and adoption, and best practices based on practitioner experience.
What industry or industries do you primarily work with?
Sustainable AI value requires a shift from defensive tool rollouts focused solely on individual tasks to offensive strategies that fundamentally redesign business models. According to research compiled from the Catalant Practice Community, many organizations fall into the trap of prioritizing short-term competitive optics over core business strategy. Market leaders avoid this pitfall by ensuring that AI implementation is explicitly driven by deep-seated business pain points and operational bottlenecks rather than a desire to experiment with novel technical feature sets.
The data suggests that while functional heads are uniquely positioned to drive localized productivity gains, they often lack the cross-departmental authority required to break down data silos and mandate enterprise-wide operational changes. Catalant field surveys indicate that tight cross-functional orchestration led by the CEO is a critical prerequisite for scaling beyond isolated test environments. Without senior executive sponsorship to align cross-departmental vision and strategy, local innovations remain fragmented, resulting in redundant investments that fail to move the needle on overall corporate performance.
Experienced consultants report that successful AI is 70% people and process transformation and only 30% technology. Insights from Catalant implementation consultants show that human change management and workforce enablement are the single most critical determinants of project survival. To ensure long-term adoption and prevent organizational friction, leaders must prioritize building cross-functional teams with diverse skillsets, establishing iterative test-and-learn loops, and embedding engaged, attentive humans-in-the-loop for rigorous oversight of model outputs.
Time saved through automation only impacts the bottom line if management has deliberately planned how to redeploy that newly freed organizational capacity. As established by Catalant strategy experts, tracking time saved or error reduction is an incomplete metric of financial return. If employees utilize their extra time on low-value tasks rather than higher-value outputs or strategic growth initiatives, the AI implementation quietly becomes a costly software rollout rather than a tangible value story that impacts corporate profitability.
Smart enterprises mitigate the risk of runaway operational overhead by embedding real-time cost and utilization telemetry directly into their AI agent and workflow orchestration architecture. Rather than treating advanced models as standard operational overhead, use frameworks that dynamically optimize for unit economics, such as cost per inference or cost per task, at runtime. This granular transparency allows teams to link variable compute and token expenditures directly back to defined business metrics, preventing capital destruction on underperforming systems.
While pilots thrive in clean sandbox environments, moving to live production exposes legacy data fragmentation and poor data quality that cripples model accuracy. The classic rule of “garbage in, garbage out” is magnified exponentially within LLMs. Moving from pilot to production frequently reveals that organizations lack a cohesive data strategy, uniform governance frameworks, and clean data pipelines. Establishing a standardized and unified data foundation is therefore a mandatory prerequisite to maintaining deployment momentum.
The fast path to overcoming the competitive deficit for unicorn talent is the targeted injection of elite independent practitioners who pair deep technical mastery with proven operating experience. While internal upskilling is necessary, the execution lifecycle cannot wait for long-term capabilities to mature. Deploying change champions and independent subject-matter experts to lead live executions provides immediate surge capacity. This hybrid model allows internal teams to acquire critical operational skills through hands-on coaching during active rollouts rather than passive offline training.
Comments are closed.