Your AI Agents Need A Chief Operating Officer, Too
As companies increasingly adopt agentic AI, the operational challenges of managing these systems like human employees are becoming critical, requiring clear rules and accountability.
Intelligence analysis by Gemini 2.5 Flash
The article argues that AI agents, especially in regulated industries, must be managed with the same rigor as human staff, including defined scopes, written rules, and assigned managers. This approach is crucial for scaling AI effectively and avoiding significant risks and project failures.
Imagine you have a super smart robot helper that can do lots of jobs, like sending important letters or checking if things are correct. Just like a new person joining your team, this robot needs clear instructions, someone to check its work, and a boss who is responsible if it makes a mistake. If you don't give it a boss and clear rules, it might do something wrong, and then no one knows who to blame, which can cause big problems.
Analysis
The rapid proliferation of agentic AI within enterprises presents a new frontier of operational challenges, moving beyond mere technical implementation to demand robust managerial oversight. The author, Rick Folgmann, posits that the core issue isn't the AI's capability, but rather its governance, advocating for a paradigm shift where AI agents are treated akin to human employees in regulated roles. This perspective is particularly salient for startups, which often prioritize rapid development over comprehensive operational frameworks, potentially exposing them to unforeseen risks as AI systems gain more autonomy.
Get Covered
At Get Covered, the author's company, the experience of deploying 16 AI agents across millions of rental units highlighted that technical hurdles were less daunting than defining reporting structures and accountability. This real-world application underscores the necessity of establishing clear lines of responsibility for each AI agent, mirroring how human teams or budgets are managed. The company's approach involves assigning a specific individual as the owner for each agent, responsible for reviewing error logs, adjusting autonomy levels, and ultimately answering for its performance. This structure prevents the 'model did it' excuse, which is unacceptable to customers and regulators alike, and ensures that AI actions are traceable and justifiable.
Deloitte’s 2026 State of AI survey
Deloitte’s 2026 State of AI survey indicates a significant trend: 23% of companies currently use agentic AI, with a projected jump to 74% within two years. This rapid adoption rate, however, is juxtaposed with the operational unpreparedness of many organizations. The article suggests that this gap necessitates a structured approach to AI management, akin to managing human staff in regulated environments. This includes establishing written rules, defining clear scopes of work, assigning managers, and conducting regular performance reviews for AI agents. Such a framework is not merely about compliance but about enabling effective scaling and ensuring that AI deployments contribute positively to the bottom line, rather than becoming liabilities.
Gartner’s cancellation forecast
Gartner's prediction that over 40% of agentic AI projects will be canceled by the end of 2027 serves as a stark warning regarding the current state of AI governance. The article attributes these anticipated failures not to technological shortcomings, but to management deficiencies, specifically the lack of defined value propositions and inadequate risk control. This highlights a critical need for decision-makers to proactively address the managerial aspects of AI deployment. By treating each AI agent as a new employee in a regulated role—complete with documented rules, limited scope, and a human manager—companies can build the necessary evidence and justification to gradually increase agent responsibility, thereby mitigating risks and maximizing the potential of their AI investments.
Key points
- AI agents require management similar to human employees, especially in regulated environments.
- Defining clear rules, scope, and assigning a human owner for each AI agent is crucial for accountability.
- Companies should start AI agent deployment with tasks that have existing rulebooks for easier auditing and validation.
- The level of autonomy for an AI agent should be configured based on the workflow's risk and detection time for errors.
- Lack of defined value or controlled risk is a primary reason for agentic AI project failures, according to Gartner's forecast.
By treating AI agents like human employees with clear rules and accountability, companies can effectively scale their AI initiatives, reduce operational risks, and achieve significant bottom-line returns. This structured approach fosters trust in AI systems and enables businesses to confidently expand their use of automation.
Companies that fail to establish clear governance and accountability for their AI agents risk project cancellations, legal penalties, and financial losses. Without proper oversight, AI systems could make unapproved decisions, leading to reputational damage and a failure to realize the promised benefits of artificial intelligence.
