Via Harvard Business Review

“The most effective [agent managers] emerged from roles already accountable for service quality, customer outcomes, and operational judgment. These individuals brought deep domain expertise and a lived understanding of what ‘good’ looks like in real customer interactions, capabilities that proved more necessary than formal AI credentials.

Organizations that developed the role successfully treated the role as an apprenticeship, immersing managers in live operations, failure reviews, and iterative test–deploy–learn cycles, while clarifying decision rights and escalation paths early. Those that centralized agent management entirely within IT or indexed on AI credentials often saw agent managers function technically while failing strategically.”

— To Thrive in the AI Era, Companies Need Agent Managers, Harvard Business Review

---

Suraj Srinivasan, who chairs Harvard’s Digital Value Lab, and Vivienne Wei, COO of Salesforce’s United Agentforce Platform, argue that the best AI Agent Managers come from backgrounds steeped in domain knowledge rather than AI expertise.

Some firms call this an AI Operator role; others describe the work as AI Orchestration. Regardless of title, Srinivasan and Wei say success in this role requires six capabilities. A cursory look at recent job postings shows that companies further along the AI Adoption curve are actively hiring for these skills.

The table below maps each capability to actual language from recent OpenAI and Salesforce job postings.

Capability (from HBR)	OpenAI: User Safety & Risk Ops Manager, Ops Enablement & Analytics	Salesforce: Agentforce/AI Deployment Strategist
1. AI Operational Literacy. Understand how agents operate, how prompts drive outcomes, and how to diagnose system failures	Partner with Product and Engineering to evolve internal operational tooling—including labeling workflows, detection and routing pipelines, classifier feedback loops, and case management interfaces; Analyze trends, bottlenecks, classifier signal strength, and operational risks	Lead analysis and strategic design of intelligent AI-powered agents leveraging Agentforce, Data Cloud, Flow, and Salesforce APIs; Experience leading and overseeing deployments of AI/LLM technologies including end-to-end engagement management
2. Functional Depth. Deep knowledge of the business process the agent supports, whether customer service, finance, or logistics	Design and scale core operational workflows, routing systems, triage logic, escalation paths, and queue architectures across all safety verticals; Build and maintain SOPs, reviewer guidelines, QA frameworks, training materials, golden sets, and structured processes	Deeply understand customers’ most complex business problems; Facilitate agent design workshops and jobs-to-be-done analysis to craft solutions; Synthesize complex business challenges into clear solution architectures
3. Systems Thinking. Visualize how agents interact across workflows, departments, and even other agents to achieve “multi-agent orchestration”	Anticipate future operational needs and design systems that scale with global growth and increasing risk complexity; Translate ambiguous or evolving product/policy requirements into clear, scalable operational processes	Provide comprehensive oversight for Agentforce deployments ensuring seamless integration with existing customer infrastructure; Monitor progress, identify roadblocks, and guide optimization for long-term reliability, scalability, and security
4. Change Resilience. Adapt quickly to shifting models and business needs, refining agent logic in weekly “test-deploy-learn” cycles	Thrive in ambiguous, high-stakes environments and balance strategy with hands-on execution; Drive operational maturity by introducing structure, documentation, measurement, and continuous improvement frameworks	Exceptional ability to confront open-ended problems in unstructured, ambiguous environments; Assess customer organizational readiness for AI adoption and guide them through necessary process transformations and reengineering
5. Prompt Craftsmanship. Excel at designing and refining the language and logic that shape agent behavior	Build and maintain SOPs, reviewer guidelines, QA frameworks, training materials, and golden sets; Partner with Product and Engineering to evolve internal tooling including labeling workflows and classifier feedback loops	Demonstrated AI consulting abilities in conversation design, responsible AI practices, and prompt engineering; Apply foundational Conversation Design skills to guide creation of intuitive conversational AI experiences
6. Designing Work Across Machines and Humans. Create AI-human hybrid workflows and motivate the human workforce in hybrid work contexts	Lead automation-first operations by identifying opportunities to reduce manual work and improve reviewer and vendor efficiency; Design and scale core operational workflows including escalation paths	Drive adoption of Responsible AI principles including risk mitigation, bias and toxicity testing, and guardrail setup; Address organizational readiness for AI adoption and guide customers through process transformations to maximize value of intelligent agents

AI-forward companies see the “Agent Manager” as an operations leadership role that happens to involve AI—not the other way around. Domain expertise is the headline requirement. AI literacy is a line item.

Srinivasan and Wei warn that when companies centralize this role in IT or over-index on AI credentials, they end up with managers who “function technically while failing strategically.” The job descriptions from OpenAI and Salesforce suggest the most AI-forward companies already hire this way.