Effective AI Leaders Start With Workflows, Not Org Charts
Old vs New: Traditional org structures constrain innovation; a workflow-first approach offers a more effective model.
AI Impact: AI requires a shift in how roles are defined, moving from structure-first to workflow-first.
Leadership Challenge: Leaders must adapt to shifting demands of AI by embracing ambiguity and reassessing old structures.
For decades, the logic of organizational design has followed the same sequence. You draw the org chart. You define the roles. Then, once people are in place, you figure out how the work actually flows between them.
It's a structure-first approach that has shaped how companies hire, promote, reorganize, and grow. And for a long time, it worked well enough that nobody questioned the order of operations.
AI is forcing that question.
Keep Reading—and Keep Leading Smarter
Create a free account to finish this piece and join a community of forward-thinking leaders unlocking tools, playbooks, and insights for thriving in the age of AI.
Have an account? Log In
Because when you introduce tools that can handle meaningful portions of knowledge work, the org chart stops being a useful starting point. Roles designed before anyone understood the workflow become containers for tasks that may no longer need a human, or may need a very different kind of human than the one who was hired. The structure you built becomes a constraint on the very transformation you're trying to achieve.
The smarter move, and one that a growing number of leaders are making, is to flip the sequence entirely. Start with workflows. Understand how work actually moves through the organization, where the bottlenecks are, where judgment matters most, and where AI can take over without degrading quality. Then design roles around what's left and figure out the reporting lines.
The Flip in Practice
Tracy Coté, Chief People Officer at Slickdeals, has been operating this way for the past year.
Historically, we'd design the org chart, define roles, then optimize workflows," she said. "AI flips that. We're now starting with workflows and outcomes, then designing roles around what's actually needed when AI should be able to handle part of the work.
This sounds like a subtle shift, but in practice it changes everything about how leaders make organizational decisions. Instead of asking "who do we need to hire?" the first question becomes "what does this work actually require?"
And the answer to that question keeps changing as AI capabilities evolve, which is why Coté also pushes back on the assumption that leadership means having the answers.
AI increases ambiguity because the tools and possibilities change weekly. Leadership is now more about setting guardrails, building literacy, and helping teams experiment responsibly without losing focus.
That ambiguity is uncomfortable for executives who built their careers on knowing the right structure for a given problem. But the discomfort is the point. A fixed structure assumes stable work, and very little about how AI affects work is stable right now.
Domenico Gagliardi, founder of the AI agent platform Kortix.ai, has taken this logic to its furthest conclusion. Rather than retrofitting an existing company, he built one from scratch around the workflow-first principle.
"We map workflows first, then design roles around what humans uniquely add," he says. The result is an organization that operates with 70% fewer people than comparable startups, not because it eliminated jobs for the sake of efficiency, but because the roles that exist were designed after the work was understood. That's a fundamentally different starting point than most companies use when they hire.
Dhanaraj S, an HR professional working across manufacturing, retail, and services in India, arrived at a similar conclusion through a more specific lens. When he set out to improve hiring at his organization, the obvious move would have been to add AI screening to the existing process.
Faster resume review, automated scheduling, maybe a chatbot for candidate questions. Layer it on top of what already exists.
Instead, he started with workflow redesign. He broke job descriptions into skill clusters and risk indicators using AI, had anonymized resumes assessed for skill alignment and compensation risk, and then generated tailored interview questions based on what the analysis surfaced.
The AI ran before human judgment entered the process, not after. The entire sequence of how a hiring decision got made was redesigned from the ground up.
The result wasn't just faster hiring. It was better hiring conversations.
More structured interviews, reduced bias, clearer hiring decisions, and fewer late-stage rejections. None of that would have happened if we’d simply plugged AI into the existing workflow. The old workflow wasn’t designed for the kind of input AI can provide. Redesigning it first was what made the technology useful.
Old Processes Can't Support New Technology
Lydia Wu, creator of the HR technology consultancy Oops, Did I Think That Out Loud, makes an even more foundational argument. She's reviewed over 800 HR tech solutions and watched countless organizations try to modernize by layering new tools on old processes. Her conclusion is direct: "We are still looking for AI to create value on top of antiquated processes that were designed for the Internet era."
That phrase, "antiquated processes designed for the Internet era," should land hard for any leader who has been in their role for more than five years. The workflows most organizations run on today were built for a world where software handled transactions and humans handled everything else. AI changes that ratio dramatically, and the processes weren't designed for it.
Wu says companies are running into a wall when they try to scale AI beyond proof of concept, and the reason is structural.
"The technology was never designed for those old processes," she argues. An AI-era transformation requires going back to the core of how the business operates, redesigning workflows at their foundation, and then applying AI on top.
This is where the workflow-first approach becomes more than a design preference. It becomes a prerequisite for getting any real value from AI at all.
And there's a second-order consequence that leaders haven't fully reckoned with. When you redesign workflows before defining roles, you often discover that the roles you need look different than the ones you have. Some tasks disappear. Others merge. New ones emerge that don't map neatly to any existing job title, particularly when AI takes over entry-level tasks.
The organizational design that made sense six months ago may not make sense anymore, and clinging to it means you're organizing people around work that no longer exists in the form you designed for.
This is why the old sequence, structure first, roles second, workflows third, produces so much friction during AI adoption. Leaders buy the tools, hand them to teams organized around pre-AI workflows, and then wonder why adoption stalls or produces marginal results. The answer is that the organizational design itself is blocking the transformation. You can't get AI-era output from an Internet-era structure.
None of this means org charts are irrelevant. Reporting lines, accountability, and spans of control still matter. But they should be the last thing you design, not the first.
The workflow tells you what the work requires. The role tells you what kind of person can do that work well. The org chart tells you how those people coordinate.
When you do it in that order, structure serves the work. When you do it in the traditional order, the work bends to fit the structure, and that bending is where value leaks out.
What This Means for Your Next Reorg
For CHROs and COOs reading this, the practical implication is uncomfortable but clear. Before your next reorg, before you fill that open headcount, before you restructure a department to "integrate AI," go map the workflows. Understand what's actually happening at the task level. Find out where decisions are being made, who's making them, and which parts of the process could be handled differently with the tools now available.
Workflow redesign before org redesign. That's the sequence. You'll almost certainly end up somewhere different than where the org chart would have taken you.