Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

How to structure AI-first organizations

Three tips for creating successful org charts

[Photo: Getty Images]

Profile photo of Dux Raymond Sy

Chief Transformation Officer · AvePoint

We’ve all heard the slogan: “AI’s great—but it’s just another tool.”

I was talking to a technology leader after a workshop recently, and they recited this line almost without thinking. Then they paused and added, “I don’t think I believe that anymore.” That admission stuck with me.

For years, the idea that artificial intelligence is “just another tool” was everywhere. I used to hear it often from skeptics, but now I hear it less and less. The shift points to something deeper in how we’re starting to use AI in the workplace.

I’ve been seeing the same pattern across many different organizations: With AI now advanced and accessible enough to produce human-like output autonomously and at scale, AI tools are reshaping work in real time. Most organizational charts haven’t caught up.

We used to look for workflows that could absorb AI. But now, we’re in a different place. A place where—in order to maximize the value of AI tools—we need to design workflows around AI.

There are three principles I keep coming back to when thinking about what an AI-first org chart looks like in practice.

1. REDESIGN AROUND DECISION RIGHTS, NOT JOB TITLES

Traditional org charts tell you who reports to whom. But the autonomy of agentic AI changes how that model holds up in the real world.

Instead of just outlining reporting lines, AI-first org charts need to do something more practical: Clarify who decides, who reviews, and who steps in when an AI agent takes action that requires a human response. When an agent can draft, analyze, route, and even initiate actions without direct supervision, it can create serious problems and ambiguity in areas of responsibility.

Organizations that are getting this right are clear about which decisions stay human-led, which are machine-assisted, and which require escalation. That clarity matters most in high-consequence areas like finance, security, legal review, and customer commitments.

The question I hear less often now is whether teams have access to AI. The more important one is whether accountability still holds once AI is embedded into the workflow.

In an AI-first organization, management layers start to look a lot more like control layers.

2. ORGANIZE TEAMS AROUND WORKFLOWS RATHER THAN FUNCTIONS

AI tends to create the most value when it reduces friction across an entire process—not when it simply makes one function faster in isolation.

I’ve noticed that organizations see very different results depending on how they approach this. The ones that move faster usually rethink accountability around how work actually flows. They clarify ownership, handoffs, and governance early—rather than treating AI like an add-on productivity layer.

That’s where the traditional functional model—marketing here, operations there, IT somewhere else—starts to show its limits. Cross-functional workflows need clear owners, shared metrics, and explicit human-AI agent handoffs. This is also where many teams start to struggle.

When no one owns the full flow of work, AI can speed up individual tasks while making the overall process harder to manage. One team may introduce an agent to save time. Another inherits the outputs without understanding how they were produced. A third ends up dealing with exceptions or risk.

What I’ve seen is that workflow-based structures reduce that fragmentation for frontier firms, where AI is at the fore. These structures make it easier to assign accountability across handoffs, define where human review matters most, and ensure governance follows the work.

3. CREATE STRUCTURES THAT SUPPORT GOVERNANCE AND CONTROL

Confidence and control often get conflated. According to AvePoint’s research, 82.7% of organizations say they’re confident or extremely confident in their ability to prevent unauthorized data access. But even among the most confident, 62% still experienced AI-related unauthorized access incidents.

Confidence doesn’t automatically translate into control. If anything, it can sometimes mask where the real gaps are. Organizations that are using AI agents and assistants effectively tend to think differently about control.

Traditional org charts can give the illusion of control, but they weren’t built to account for how AI systems operate. What I’ve seen work instead is a more deliberate approach: Assigning clear ownership for high-risk workflows, defining escalation paths for AI-driven exceptions, centralizing policy oversight, and embedding governance closer to functions like security, legal, and operations, where decisions are actually being made.

When organizations redraw the org chart around control points, workflow ownership, and governance accountability, technology and authority stop operating separately, and start working in tandem, toward similar goals. That’s when AI can scale with less friction, and a lot more trust.


THE FRONTIER FIRM

The AI-first firm—what Microsoft has termed the “Frontier Firm”—is not defined by how many AI tools it deploys, but by whether it can govern and use AI well. That means redrawing the org chart around control points, workflow ownership, and governance accountability, so technology does not sit apart from authority. When organizations do that well, governance technology becomes more effective and agentic AI can scale with less friction and risk.

AI amplifies productivity, but without the right scaffolding, it also amplifies risk. That’s why it’s important to give AI tools the correct kinds of support.

Dux Raymond Sy is chief transformation officer for AvePoint.