Thousands of AI Agents: 13 C-Level Leaders With New Responsibilities

At the start of this year, many businesses were experimenting with AI agents and had only a handful deployed to production. Leaders were beginning to learn what AI agents could do, who should develop them, how to determine whether they’re production-ready, and what steps were needed to grow employee adoption.

Dozens to Thousands of AI Agents: The 13 C-Level Leaders With New Responsibilities

Now, some organizations are facing the next challenge as they grow from dozens to thousands of AI agents in production. Many are trying to keep AI governance up to speed with this scale, while others are working through proliferating shadow AI challenges. Some already see the need for AI agent orchestration platforms as they connect task-based agents into role-based assistance and re-engineered business processes.

At last week’s Coffee With Digital Trailblazers, a LinkedIn Live event I host weekly at 11am ET, we discus

sed the governance challenges in the episode, Vibe, Buy, or Automate? Deciding How to Build AI Agents.

With Gartner predicting that enterprises will manage an average of 150,000 agents in two years, my concern is that organizations will repeat the mistakes made in the past – with a magnitude of a mess somewhere between the 650+ SaaS platforms and millions of spreadsheets enterprises manage. Listen to the episode and review its research slide for details.

The leaders will drive AI outcomes and minimize risks

Whose leadership responsibility is it to scale from dozens to thousands of agents, ensure they are delivering value, realign employee roles, and minimize risks? Here’s a breakdown of at least one new responsibility for 13 C-Level Leaders.

CIO: Redesign IT’s digital operating model

Last year, I wrote about what world-class IT looks like in the AI era and launched workshops to guide departments through the transformation. I’ve advised on revisiting Agile for the AI era and shifting from running AI experiments to delivering value.

The digital operating model that enabled top CIOs to deliver cloud, mobile, and analytics innovations needs to be redesigned to drive AI’s transformation. For organizations looking to do more than reshape business and drive transformation, IT is more important than ever, but the CIO must lead the department’s reorganization in skills, roles, processes, and platforms.

“When automakers moved from mass production to flexible manufacturing, the winners did not just buy better machines,” says Sheldon Monteiro, chief product officer at Publicis Sapient. “They changed the system around them: skills, incentives, information flows, and who makes decisions. We’re about to relearn that lesson with AI.”

CDO: Establish dynamic data governance for AI agents

A new survey from Collibra by The Harris Poll reports that 72% of decision-makers agree that when their organization’s AI initiatives fall short, the root cause almost always traces back to an unaligned or poor data foundation.

Data governance in the pre-AI era focused on structured data, compliance, policies, data quality, ownership, and access privileges. In the AI era, unstructured data governance matters far more, and chief data officers (CDOs) should lead knowledge management and shape AI’s context layer.

“For years, CDOs built data governance for humans, but now they have to build it for agents that can access data, make decisions, and take action on their own,” says Felix Van de Maele, CEO and founder of Collibra. “Static policies are not enough for that world, and agents need governed context to make the right decisions and have controls that can be enforced while they are operating. The CDO’s role is shifting from defining how data should be used to making sure those rules actually hold at runtime.”

CISO: Govern agent identity and authorizations

Identity, access management, and threat detection all require elevated security strategies when AI agents outnumber employees and contractors. CISOs should work with their CIOs to converge their NOCs and SOCs so that security specialists can focus on AI’s growing security challenges.

Jacob DePriest, CISO and CIO at 1Password, says, “As agents move into production, CIOs and CISOs need to establish controls that make an agent’s actions auditable to give leaders clear visibility into the systems and data it’s accessing.”

CTrO: Oversee change management and value delivery

Not every organization has a chief transformation officer (CtrO) or an agile PMO. But someone needs to oversee value delivery, ensure AI agents have accountable owners, and lead change management efforts. A change process brings these three factors together into an agile delivery model that evolves as agents are deployed and updated.

“The leadership skill that matters isn’t picking the right orchestration platform; it’s knowing how much agentic autonomy your organization can actually verify,” says Barr Moses, co-founder and CEO at Monte Carlo. “Going from a dozen agents to a thousand doesn’t just multiply the workflows; it multiplies the number of potential failures, and so expands the landscape of required visibility. Leaders who scale successfully treat trust as a gate, not an afterthought. An agent earns broader authority only after its behavior has been observed and proven.”

CMO: Reimagine the brand and drive growth opportunities

Much of the focus in 2026 has been on AI delivering productivity and workflow efficiencies. I predicted that in 2026, we’d see AI’s Airbnb or Uber moment, but AI agents embedded in products and customer-facing capabilities haven’t materialized at scale. It’s coming, and a key role for CMOs is to align the organization on AI’s brand promise and customer opportunities.

“Brands have to be able to stand out in a sea of agents,” says Emily Ketchen, SVP and CMO, intelligent devices group and international markets at Lenovo, on how AI agents are reshaping the CMO’s job. “When I think about the role of a chief marketing officer today, we are really responsible for thinking about how to inspire teams to be able to lean into brand-new technologies to bring to life what those mean for customers.”

CCO: Establish KPIs for human and AI agents

A top area for growth and CSAT improvements is the contact center. Scoring ROI in CX requires the chief customer officer (CCO) to balance responsibilities between human and AI agents, then select KPIs to align operations.

“Agent orchestration means AI and humans making decisions side by side, not AI agents running the show alone,” says Jaime Meritt, chief product officer at Verint. “As that scales from dozens of task bots to thousands of coordinated agents, the job shifts from managing tools to governing a workforce, proving outcomes instead of chasing pilots. The leaders who get this right can answer two critical questions whenever an agent affects a customer: who reviewed the decision, and how quickly was a mistake caught?”

COO: Align performance objectives for the human + AI workforce

What does the new org model look like, and how is performance managed when operations is orchestrated between AI agents and people? Some say CIOs and CHROs will be responsible for realigning the organization. But in large operations, especially in manufacturing, construction, and field services, the COO should lead how people, AI agents, and operations will coordinate around quality, cost, and safety.

“As organizations scale from dozens to thousands of AI agents, leaders should think of them as a highly capable but extremely naive workforce,” says Nirmal Mukhi, chief architect at ASAPP. “Agents need clearly defined roles, permissions, supervision, and performance management, with human teams ultimately accountable for their outcomes. The organizations that succeed won’t be those running the most agents or consuming the most tokens, but those that teach their people how to manage a human+AI workforce effectively.”

CAIO: Re-engineering business processes with AI agents

Not all organizations have a Chief AI Officer, and IMHO, many of them shouldn’t. But when I think of digital-first or AI-native companies with significant data, IP, and analytics capabilities, someone has to lead agentic workflow design. This role likely falls to the CIO for organizations that don’t have a CAIO.

“Just as you wouldn’t build a human organization without defining roles, skills, access, and accountability, you can’t build an agentic organization that way either,” says Joseph Kim, CEO at Druid AI. “The new leadership skill is orchestration: knowing which agents should do what, what capabilities and permissions they need, which agents can direct others, and where humans need to remain accountable. We’re effectively designing a new organizational structure—one where the workforce includes both AI agents and humans.”

CHRO mandate: Redefine people’s roles and job descriptions for the AI era

Years before genAI and AI agents, I advised leaders to avoid using the word automation. It screams “danger” to employees who fear for their jobs without knowing their future opportunities at the company or how to develop their careers. CHROs must lead lifelong learning programs and ensure C-level leaders define the skills, job responsibilities, and development programs for their managers and employees.

“As AI transforms the workplace, our opportunity as HR leaders is to look deeper into the work itself – breaking down roles into individual tasks to determine where human judgment matters most and what an agent can do,” said Ashley Goldsmith, chief people officer, Workday. “When we understand how work truly gets done, we can be far more intentional about upskilling and redeploying talent to make the greatest impact.”

CTO mandate: Define build and lifecycle standards

CTOs and architects have the challenge of creating standards and lifecycle management for AI agents, whether they are bought through SaaS platforms, vibe’d using development tools, or built by development teams. How do you ensure the AI agents meet development standards? How are agents cataloged and reused before enterprises end up with duplicate agents making inconsistent decisions? CIOs, CTOs, architects, and lead developers have many questions and responsibilities to own when scaling to thousands of AI agents.

“The jump from dozens to thousands of agents, all connecting through orchestration platforms, is where careful planning of the agentic architecture makes a big difference,” says Phillip Merrick, co-founder and CEO at pgEdge. “Unix philosophy got here first: instead of one agent doing everything, break the work into steps and assign each step its own agent. When you use this micro-distributed approach, each agent in the pipeline acts as a checkpoint that ensures quality control and standards are applied. The end result is achieved more effectively, with a more accurate result – exactly what is needed when orchestrating at scale.”

CRO: Ensure agent observability and auditability

Regulated enterprises often have chief risk officers (CROs); otherwise, responsibilities are often shared between CISOs, CFOs, COOs, IT ops leads, and heads of legal. When deploying AI agents: (1) an owner must be assigned, (2) the owner accepts responsibility for the agent’s decisions, (3) the implementation must follow observability and auditing standards, (4) agenticops is implemented to monitor runtime behaviors.

Rishi Bhargava, co-founder and CRO at Descope, says risk management leaders have to consider accountability at scale. “At a thousand agents, auditors start asking hard questions about who authorized them. Every agent action should trace back to a specific person or team that delegated the authority, with the scope of that authority defined before the agent ever runs,” Bhargava says.

CFO: Define trust and own tokenomics

I have two key responsibilities for CFOs. First, they are best positioned to define the criteria for trusting AI agents, specifically, what needs to be explainable, auditable, and defensible to regulators, along with the AI agent owner’s accountabilities.

Second, some define tokenomics tactically, as in, optimizing token spend and AI model selection. But I prefer the FinOps foundation’s definition: “AI Tokenomics is the study of how the production, distribution, and consumption of AI tokens generate business outcomes and AI value within an organization.” This definition covers both cost and value responsibilities and should be a key CFO responsibility, partnering with the CIO to avoid AI cost debt.

“The CFO’s first question is not what an agent can do, but what it is authorized to do and who remains accountable for the outcome,” says Manoj Swaminathan, president and chief product officer, Autonomous Suite, SAP, and member of the Extended Board. “As AI agents take on more operational work, finance leaders must ensure every action is governed, auditable, and traceable within the same risk and compliance framework that underpins the business. Trust in the numbers relies on the ability to trace agent decisions clearly. Ultimately, the expectations for agents are the same as for anyone trusted with a company’s capital.”

CEO: Align C-level leaders on a transformation imperative

Many, if not all, of the responsibilities I’ve outlined are important in most organizations. But few businesses have C-level leaders overseeing all the functions I’ve outlined. The CEO’s first mandate is to ensure leadership responsibilities are assigned and clearly defined. Expect conflicts as AI impacts more operational functions and potentially disrupts aspects of the pre-AI business model.

Which is why CEOs must be involved in restating the organization’s mission, ethics, and priorities – and update this frequently, no less than twice per year, given the rapid evolution of AI’s opportunities and risks.

“AI will either be the greatest equalizer ever invented, or the worst source of injustice. The challenge is monumental. Even under the best circumstances, the transition to this new AI era will be one of the most turbulent times in human history. […] I don’t see evidence that leaders, experts, and communities are confronting the challenges adequately.” – Bill Gates in The turbulent AI era is here. The choices we make now are critical.