The AI Org Chart: The Roles Forming Under the Chief AI Officer

In Part 1 of The AI Org Chart, we looked at the seat at the top: the Chief AI Officer. Three quarters of the organizations in IBM's 2026 CEO study now have one.¹

The obvious next question is: who reports to them?

I want to be careful and honest here, because this is where a lot of AI content turns into fantasy. So this article has two halves. First, the AI titles that already exist, with real hiring data behind them. Second, my model of where org charts are heading: five layers I believe will form under the CAIO. Those are my proposed titles. They are not in a census yet.

Which AI job titles already exist today?

Start with what is real and measured.

Chief AI Officer

The top seat. IBM found 76% of surveyed organizations had a CAIO in 2026, up from 26% in 2025.¹ U.S. federal agencies were required to retain or designate one under OMB Memorandum M-25-21, and the largest agencies must make their CAIO the vice-chair of an AI governance board chaired at the Deputy Secretary level.⁵

Head of AI and Director of AI

These are real titles with hiring data. LinkedIn's research on 2025 U.S. AI hiring, as reported by IT Brief, broke out hires for Head of AI, Director of AI and Member of Technical Staff.³ Back in 2023, The National CIO Review reported a 2.6-fold increase in companies creating a Head of AI role over the previous five years.⁴

AI engineer, and AI consultant or strategist

LinkedIn's 2026 Jobs on the Rise list put AI engineers first and AI consultants and strategists second among the fastest-growing roles in the U.S., based on millions of jobs started by LinkedIn members.² Notice what the second one is. Not builders. People who help organizations plan and adopt AI. That tells you where the pressure is.

Why do I think the org chart will go further?

Because building AI and running AI are two different jobs, and most org charts only have seats for the first one.

In the same IBM study, CEOs said only 25% of their workforce uses AI regularly in their jobs.¹ Companies bought the tools and hired engineers. What is missing is the operating layer: people who run AI inside marketing, sales and delivery every day, the way a COO runs operations.

What layers do I see forming under the CAIO?

President of AI Operations

The operating chief for AI. Where the CAIO sets strategy and governance, this person makes sure the agents actually run the business day to day: workflows live, outputs reviewed, problems fixed. Think COO, but the team includes agents.

Head of AI Acquisition

AI across marketing and sales. Ad creative variation, inbox agents, lead qualification, follow up, tracking and reporting. This is the seat closest to revenue, and in an expert business it is usually the first one that pays for itself.

Senior AI Operator

The builder who knows the business. A Senior AI Operator can take a brief and direct agents to produce a whole system: pages, emails, workflows, ads. They also hold the quality standard and know when an output is wrong even if it looks polished.

AI Agent Manager

Owns a group of agents the way a manager owns a team. Writes and updates their instructions, maintains their knowledge, sets their permissions, monitors their work and fixes them when they drift. This role barely exists by name yet. I believe it will be one of the most common jobs on the chart.

Associate AI Operator

The entry seat. Reviews outputs against the standard, handles exceptions the agents escalate, and learns the business by watching the system run. This is how the next Senior AI Operators get trained.

How do the layers fit together?

Here is the chart, top to bottom, with the real titles and my proposed layers side by side.

What skills matter for every layer?

Whether you hire these roles or hold them yourself, the same four skills show up at every level.

• Writing a clear brief. Agents do exactly what you describe. Vague in, vague out.
• Defining the standard. What does good look like for this page, this email, this reply? Write it down.
• Reviewing fast. The bottleneck moves from making the work to checking it. Review is the new production. My five-minute process is in A Simple QA Process for AI Output.
• Permission discipline. What each agent can read, write, send or spend, and who approves (the spending rules are in Agents That Spend Money).

None of these are technical skills in the engineering sense. They are management skills applied to a new kind of worker. That is good news for experts and operators. If you have ever trained a team member, written an SOP or reviewed someone else's work before it went to a client, you already have the foundation for the operating layers of this chart.

What does this mean for an expert business?

You will not hire seven people. You probably should not hire any. But you will need every one of these jobs done.

In a small business, these layers become hats and agents. The founder holds the CAIO seat. Agents fill most of the operating seats, and the founder, or one trusted operator, holds the review and judgment layers. That is how my own business has run since 2024, without a single hire.

That is exactly what Part 3 is about: how a one-person business staffs this whole org chart with agents.

For agency owners whose people deliver client work, the next step is training them as AI operators: a system and agents installed in the business, named operators trained on it, and the first live client builds run together. And if the first seat to fill is Head of AI Acquisition, start with an inbox agent.

Keep building.Ekai