AI ROI Leaks to Individuals; Organizations Must Adapt Culture and Workflows

The ROI Leak: Why AI's Promise is Being Hoarded by Individuals, Not Captured by Organizations

The current enterprise adoption of AI is a mixed bag, earning a C-minus at best, with a select 5-15% of organizations achieving A-plus results. This isn't because AI lacks value; rather, the tangible ROI is "leaking" to individual employees who are proactively adopting these tools, often outside of official company channels. These ambitious individuals are leveraging AI to enhance their productivity, but without organizational adaptation in culture, incentives, and workflows, these gains remain siloed, failing to translate into broad organizational impact. This conversation reveals a critical hidden consequence: the potential for a widening gap between individual capability and organizational structure, where traditional org charts and rigid workflows actively impede the systemic adoption of AI. Leaders and teams looking to harness AI for genuine business transformation should read this to understand the systemic barriers to AI ROI and how to begin dismantling them, gaining a competitive advantage by aligning individual AI prowess with organizational strategy.

The Hidden Cost of "Cutting" Workflows

The prevailing narrative around AI adoption focuses on efficiency, on "cutting" existing workflows and tasks. This is understandable; organizations are built to optimize and exploit what they already do well. However, Greg Shove highlights a critical oversight: this focus on cutting, while immediately productive, often fails to address the "create" aspect--developing new, higher-value work. This leads to a situation where employees, empowered by AI, complete their existing tasks faster, but the freed-up time and energy don't automatically translate into strategic growth or innovation. Instead, this "leaked" ROI is often absorbed by individuals who might use the time for personal pursuits or simply maintain their existing performance levels with less effort, a rational response to the inherent uncertainty of AI's impact on job security.

"The ROI is being kept by the employee. I think that's very true. 10 to 15% of every organization are the growth mindset, are they ambitious, are those that are curious... and they're keeping the gains."

-- Greg Shove

This dynamic creates a subtle but significant systemic issue: the organization invests in AI, but the expected organizational-level productivity gains are siphoned off by individual efficiency improvements. This is compounded by a lack of clear organizational direction. Without a defined "why" for AI adoption, employees are left to navigate its integration themselves, leading to a fragmented approach where individual productivity can even be seen as a risk to established workflows and job roles. The consequence? Companies are essentially paying for productivity gains that aren't being reinvested into the business's future.

The "Disposable Software" Delusion

A key insight emerging from the conversation is the fundamental shift AI enables: the rise of "disposable software." Unlike traditional Software-as-a-Service (SaaS) products that are built for longevity and productization, AI allows individuals to quickly spin up custom solutions--like GPTs or automations--for specific, often one-off tasks. Shove illustrates this with his birthday party seating chart example. This capability, while incredibly powerful for immediate problem-solving, clashes directly with the traditional enterprise mindset of building and maintaining robust, long-term software assets.

"We thought that SAS software... we went from albums to singles, but now we're not even having the singles anymore. Now we're just basically writing the tunes as we need them, and then we're throwing them away."

-- Greg Shove

The consequence for organizations is a tension between empowering individual creativity and maintaining control over corporate data and workflows. CIOs express concern about "disposable software" running rampant, potentially creating security risks and data silos. This highlights a systemic challenge: how do organizations foster the agility and rapid iteration that AI enables without sacrificing stability and governance? The failure to adapt to this new paradigm means companies might miss out on the true potential of AI, clinging to outdated models of software development and deployment that are ill-suited for the current technological landscape.

The "A-Plus" Organization: Manifesto, Access, and Culture

The distinction between C-minus and A-plus organizations lies not in advanced technical prowess, but in foundational cultural and strategic elements. Shove emphasizes that truly successful AI adopters have a clear "why"--an AI manifesto--that guides their adoption. This manifesto isn't just about stating intent; it's about fostering a culture that celebrates AI exploration, encourages experimentation with even "disposable" AI tools, and explicitly communicates that using AI is not cheating. This cultural foundation is crucial for unleashing employees.

Furthermore, A-plus organizations ensure equitable access to "great AI." The anecdote of a pharmaceutical company providing a "dumb" free version of CoPilot to most employees while reserving the "smart," paid version for a select few is a stark illustration of what not to do. This disparity breeds resentment, creates a class system within the organization, and ultimately negates the potential benefits of AI by forcing high-performing teams to collaborate with those using inferior tools. The cumulative output, in such a scenario, tends to fall to the lowest common denominator.

"Why are our people stalling out... when 90% of you said we told people what they're not allowed to do and 10% of you have said I've told people what I hope they do. Is there any wonder why people are stalling out?"

-- Anonymous CEO in a diagnostic session

This points to a systemic failure in many organizations: a focus on governance and restriction ("what they're not allowed to do") over vision and enablement ("what I hope they do"). The A-plus organizations, conversely, prioritize providing access to the best tools and fostering an environment where employees are not just permitted, but actively encouraged, to leverage AI to become more effective in their roles. This requires leaders to actively shape the narrative and provide the necessary support, moving beyond mere permission to proactive enablement.

Actionable Takeaways for AI Transformation

• Develop and Communicate an AI Manifesto: Clearly articulate the "why" behind AI adoption within your organization. This should be more than a governance policy; it needs to be a bold vision that inspires and guides. (Immediate)
• Ensure Equitable Access to Quality AI Tools: Avoid creating AI tiers within your organization. Provide all employees with access to the best available AI tools, recognizing that disparate access creates systemic disadvantages. (Immediate)
• Foster a Culture of AI Exploration: Actively encourage employees to experiment with AI, including building custom GPTs or automations, even if they are "disposable." Celebrate learning and experimentation over immediate, perfect outcomes. (Ongoing)
• Redefine Workflows for "Creation," Not Just "Cutting": Managers must proactively identify new, higher-value work that can be pursued with AI-enabled efficiency gains, rather than simply expecting employees to fill freed-up time with existing tasks. (Over the next quarter)
• Carve Out Dedicated Time for Strategic Thinking: Leaders and teams need to intentionally block time--like offsites or dedicated strategy sessions--to explore AI's potential and redefine future business models, combating Parkinson's Law. (This pays off in 3-6 months)
• Invest in Use Case Discovery and Coaching: For the majority of employees who aren't early adopters, provide structured support to help them identify and implement AI use cases specific to their roles and workflows. (This pays off in 6-12 months)
• Embrace Organizational Ambidexterity: Recognize the inherent tension between exploiting existing capabilities and exploring new opportunities. Actively build structures and processes that support both, as AI makes exploitation cheaper and exploration more critical. (This pays off in 12-18 months)