Reimagine Who Does Work

With Generative AI, We Are Thinking Too Small

That is a bold claim considering artificial intelligence capital expenditure is in the trillions of dollars, hiring for many jobs appears frozen, and it is disrupting all industries from accounting to consulting to education to travel. It is ubiquitous. The reason the scope is bigger than we think is that most generative AI use cases so far have been task-oriented. You type something in, and the machine does something else. It is search on steroids.

Even business-to-business use cases, such as reducing demand for monotonous and algorithmic work, focused at the task level: individual tasks, team tasks, and department tasks. The real transformation happens when we think bigger than tasks and reimagine entire workflows 1. The shift from task automation to workflow reimagining is where structural, strategic value emerges.

Digital Employees

Forward-thinking human resources departments call agents digital employees. Digital agents can take action based on guidelines and algorithms that users have set. This is far more than answering questions and drafting emails. Remember how robotic process automation seemed revolutionary five years ago? In the old days around 2018, you wired up different applications and gave them specific if-this-then-that directions. Entire businesses were built around that model.

Agents operate at a different level. They can evaluate options and select among choices. Instead of hard programming a workflow, you have digital agents that think, decide, and improve over time. This is the leap from automation to autonomy, and it changes how organizations should think about their workforce composition and workflow design.

Reimagined Workflows

Some companies report not getting significant gains from generative AI initiatives because the application is voluntary, ad hoc, unstructured, or simply juvenile. It is like dropping a food processor in the corner of a commercial kitchen to see what happens. The real structural and strategic win comes when the work itself is reimagined, not when generative AI is a voluntary add-in to your browser.

McKinsey examined 80 different use cases. In one example, a technology company had thousands of smaller accounts not getting attention from their inside sales force. With generative AI, the company deployed four agents: one prioritized accounts using public and proprietary data, one conducted outreach to qualify them, one managed follow-up and stratified leads, and one scheduled contact with a human employee. The results were significant: 30 to 50 percent time savings and 7 to 12 percent revenue increase from these small accounts.

McKinsey notes that managers and specialists are increasingly acting as orchestrators and validators rather than executors. Domain experts such as data analysts, underwriters, and engineers partner with agents that perform initial analysis or generate draft outputs. This is not incremental change. It is a 100 percent reimagining of how work gets done.

Not Incremental Change

This takes guts. This is not a 10 percent change in how work is done. It is a 100 percent change. If you have the equivalent of magic to do mundane, repetitive, or difficult and novel work, why would you not use it? The future becomes a mix of what humans do, what agents and robots do, and how humans, agents, and robots coordinate together.

This is an example of yes-and thinking. The organizations that approach AI with curiosity rather than fear will outpace those that treat it as a threat or a passing trend. The question is not whether to adopt these tools but how to reimagine workflows so that humans, agents, and robots each do what they do best.

Courage and Leadership

From 30 years of corporate experience, one observation stands out: people do not like change. Change is tough. Companies gain massive inertia with norms, ways of working, and entitlement. Change is glacial. McKinsey offers questions leaders should ask about reimagining their business for future value, leading AI as core business transformation, building a culture of experimentation, ensuring trust and safety, equipping managers to lead teams of people and agents, and preparing workers for new skills and roles.

Most organizations would score poorly on that list. The gap between knowing what to do and actually doing it is where leadership matters. Courage is required to dismantle existing workflows, retrain staff, and accept the disruption that comes with fundamental change. The companies that wait for certainty will find that their competitors have already moved.

Diagnose Your Current Workflow

This is the most difficult and painful step. You need to provide a fair amount of detail about the many steps you take. In fact, you probably do not even recognize how much minutiae you are doing. When consultants use value stream mapping, it is time-consuming, invasive, but also enlightening 2. Even if you automate nothing, cataloging and dissecting your activities to distinguish valuable work from routine work is super useful.

Give the boring stuff to the agent. The diagnosis phase reveals opportunities you did not know existed. Most professionals are so embedded in their daily routines that they cannot see the waste. Mapping the workflow forces objectivity and creates the foundation for meaningful redesign rather than superficial automation.

Redesign the Activities

This is where the rubber meets the road. Ask what areas agents can do just as well or better than you can at pennies on the dollar. What work should you give away? Redesign the workflow so that humans focus on judgment, framing, decisions, and exceptions. AI agents handle analysis, coordination, monitoring, and first drafts. For each step, specify the human role, the AI agent role, and the checkpoints where humans stay in control.

The redesign phase is where strategy meets execution. It requires understanding both the capabilities of current AI tools and the specific needs of your workflow. The goal is not to eliminate humans but to elevate them. When agents handle the routine work, humans can focus on the judgment-intensive activities where they add the most value.

Stay Open-Minded

There is a good chance you will not automate your workflow entirely. However, it is instructive to figure out what you should do more of and what you should do less of. Analyze which current skills are most leveraged in a human-AI partnership, which skills are becoming less valuable or more automatable, and which new skills you should develop, such as AI oversight, prompt design, and sense-making. Propose a simple 60 to 90 day skill-upgrade plan.

The open-mindedness extends beyond tools to identity. If your value has been tied to performing certain tasks, and agents now perform those tasks, you must redefine your professional value. This is uncomfortable but necessary. The professionals who thrive will be those who see themselves as orchestrators and validators rather than executors of routine work.

Create a System

For those who want to go deeper, create a system with feedback loops. Track signals such as speed, quality, outcomes, errors, and user response. Have AI agents summarize patterns and surface insights. Adjust decisions or workflow steps based on those signals. Design this as a lightweight, repeatable system that evolves over time.

The system approach transforms AI from a one-time project into an ongoing capability. Feedback loops ensure that the workflow improves continuously rather than degrading after initial deployment. The organizations that build these systems will compound their advantages over those that treat AI as a set-and-forget implementation.

Fix the Process First

At Deloitte, the principle was clear: fix your process first. If you take a bad process and speed it up with technology, all you get is a faster death. This is about orchestration, not just acceleration. We are all going to have to learn how to dance with the agents 3.

The sequence matters. Diagnose first, then redesign, then implement, then build feedback systems. Skipping steps leads to automating broken processes, which amplifies existing problems at machine speed. The discipline of process improvement is the prerequisite for successful AI implementation, and it always has been.

• 1RPA and the Evolution to Agentic Automation
• 2What Is Value Stream Mapping and Why Is It Important?
• 3Value Stream Mapping: A Visual Approach to Process Optimization