Your AI Strategy Needs a Work Strategy
By Salima Hemani, CEO, SZH Consulting
There is no shortage of organizations investing in AI and doing so rapidly, from deploying copilots, experimenting with agents, to automating tasks and building AI strategies.
But what they are often overlooking is what we believe a critical risk in moving too quickly from “What can AI do?” to “Where can we deploy it?”
Implementing AI into an organization designed around yesterday’s assumptions about work may make pieces of that organization faster, without necessarily making the organization better. During our recent EXPOSED! How AI is Revealing Organizational Performance webinar with Humanico, one theme kept surfacing: AI isn't simply creating new organizational challenges. It is exposing the ones that were already there.
Unclear roles, fragmented processes, capability gaps, slow decisions, weak coordination, leadership blind spots are conditions that AI can magnify, which makes this moment an opportunity to do something bigger than adopt new technology. It is an opportunity to rethink and redesign how work gets done.
AI Is Exposing the Organization
One of the concepts we explored during the AI EXPOSED! Webinar was the perception gap, which is the distance between how leaders believe their organization operates and what is actually happening inside it. AI makes that gap increasingly consequential. An organization may believe that it has a technology problem when the real constraint is unclear decision rights. It may assume it needs new AI talent when the greater issue is skills debt, which is accrued when there is a gap between the capabilities an organization has today and the capabilities it needs to execute its strategy and operate effectively in the future. In the AI era, skills debt becomes particularly important. For example, a company may deploy AI broadly but not develop employees' ability to work effectively with AI, critically evaluate its outputs, redesign workflows, use data to make decisions, or exercise the higher-order human judgment that becomes more valuable as routine work is automated.
Organizations can rack up other forms of debt along the way, too. Trust debt shows up when employees don’t understand how decisions about AI and their work are being made. Dignity debt shows up when people are treated as units of output rather than as the ones who actually know the work best and deserve a hand in redesigning it.
Visibility, however, is only valuable if it changes what leaders do next.
Don't Just Add AI to the Work. Rethink the Work.
This is where AI strategy and organization strategy need to converge. Rather than beginning with jobs or technology, organizations should begin with the outcomes they are trying to create and examine the work required to deliver them.
Then ask four critical questions:
Where can AI perform more efficiently or consistently, with appropriate oversight?
Where can AI increase human capacity, insight, creativity, or decision quality?
Where does AI make it possible to fundamentally rethink a workflow rather than simply accelerate individual tasks?
Where do judgment, relationships, empathy, creativity, accountability, contextual understanding or leadership create differentiated value?
That last question is particularly important. The goal should not simply be determining what AI can do. Organizations should intentionally decide where humans create the greatest value.
What Might This Look Like in Practice?
Consider a biopharma company preparing for commercialization. The organization may initially see AI as an opportunity to help employees work faster, summarizing information, creating first drafts, analyzing data or automating administrative activities. Now consider a cross-functional launch decision. Information may move among Clinical, Medical, Regulatory, Commercial and Supply Chain teams. Each function conducts its own analysis, prepares materials and escalates recommendations through different leadership channels. AI could make each of those activities faster. But what if the organization asked a different question: Should the workflow still operate this way at all?
Imagine AI continuously synthesizing relevant information across functions, identifying inconsistencies, surfacing emerging risks and developing decision scenarios before leaders ever enter the room. Now the questions becomes organizational:
Who really needs to participate in the decision?
What information should leaders receive?
What work disappears?
What new capabilities become important?
And what becomes possible when highly skilled people spend less time assembling information and more time interpreting it and making decisions?
Another similar example is of a professional services firm. The traditional model often relies on junior professionals to conduct research, analyze data, synthesize findings, develop presentations and prepare senior colleagues for client conversations. AI can now perform or dramatically accelerate significant portions of that work. One response is to give every consultant an AI assistant and expect productivity to improve. A more fundamental response is to ask: If we were designing the consulting model today, would we design it this way?
If smaller teams of experienced practitioners supported by AI can accomplish work that once required much larger teams, what should the team look like? What should junior professionals learn if some of the work through which they historically developed expertise disappears? How should projects be priced if hours are increasingly disconnected from value? And where does human expertise become more important, rather than less?
The resulting model might look very different. It might be focused on smaller teams with greater senior involvement, AI-enabled intelligence, faster cycles, and more time spent on judgment, problem-solving and client interaction.
This isn’t just anecdotal. MIT’s Project NANDA found that roughly 95% of enterprise generative AI pilots are producing no measurable financial return, and pointed to organizational and workflow gaps, not the technology itself, as the real culprit (The GenAI Divide: State of AI in Business 2025, MIT Media Lab, July 2025). Harvard Business Review has made a similar case: organizations get real value from AI when they redesign the underlying process first, rather than layering new tools onto old ways of working (Wilson & Daugherty, “The Secret to Successful AI-Driven Process Redesign,” HBR, Jan/Feb 2025). And MIT Sloan Management Review’s research points to the same conclusion: AI’s biggest impact comes from how it reshapes entire workflows i.e., how tasks are sequenced, grouped, and handed between humans and machines, and not from speeding up individual tasks in isolation (“How AI Is Reshaping Workflows and Redefining Jobs,” MIT Sloan Management Review, 2025).
These examples are a reminder that AI strategy, workforce strategy, and organization design can’t stay separate conversations. When technology changes the work, the work changes the organization. And once the work changes, nothing else can stay static for long.
From AI Adoption to Organizational Reinvention
This suggests a different sequence for leaders.
Instead of:
AI Strategy → Technology → Adoption
consider:
And bring the people closest to the work into the redesign.
They’re usually the ones who understand the exceptions, workarounds, relationships, and judgment calls that never show up in a process map or job description. Their insight is often what separates work that merely looks automatable from work where context and human judgment matter enormously.
The goal was never to build the organization with the most AI in it.
It’s to build one where humans and AI together create more value than either could alone. McKinsey’s research on “superagency” in the workplace makes a similar point: the gap holding most organizations back is leadership and organizational design, not the technology itself (“Superagency in the Workplace: Empowering People to Unlock AI’s Full Potential,” McKinsey & Company, January 2025).
That takes a technology strategy, yes.
But it also takes real choices about work, structure, roles, capabilities, decision rights, governance, leadership, culture, and the human experience of all of it.
A Different Question for Leaders
As organizations build out their AI strategies, the most important question probably isn’t “Where can we use AI?”
It might be closer to:
“If we were designing how this work gets done today - knowing everything we now know, and everything AI now makes possible - would we design it this way?”
If the answer is no, adding AI to the system you already have probably won’t be enough.
At SZH Consulting, we help organizations rethink how work gets done - bringing together organization design, workforce strategy, leadership, culture and change to build organizations ready for what's next.
