---
id: "action-manage-ai-agents"
type: "action-item"
source_timestamps: ["§ Tell Us: How Are You Managing AI Agents?", "¶11"]
tags: ["management", "ai-agents"]
related: ["concept-autonomous-agentic-operations", "question-managing-agents-challenges"]
action: "Define official responsibilities for managing autonomous AI agents and taking accountability for their ultimate outputs."
outcome: "Clear organizational accountability and support structures for the shift from conversational AI to action-oriented AI."
speakers: ["Gretchen Gavett"]
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-cl-77-new-data-using-ai"
sourceUrl: "https://hbr.org/2026/06/new-data-on-how-were-really-using-ai"
sourceTitle: "New Data on How We’re Really Using AI"
---
# Formalize the management of autonomous AI agents

**Action:** Define official responsibilities for managing autonomous AI agents and for taking accountability for their ultimate outputs.
**Outcome:** Clear organizational accountability and support structures for the shift from conversational AI to action-oriented AI.

As AI transitions from advising to 'doing' ([[concept-autonomous-agentic-operations]]), organizations must stop treating AI merely as a tool and start treating it as a *managed entity*. Leaders should explicitly decide whether managing agent output is an **official part of a worker's job description**, identify the unique challenges of that oversight, and provide targeted support for the new managerial shift. HBR ([[entity-org-harvard-business-review-d8]]) is actively surveying readers on exactly this experience — the open unknowns are tracked in [[question-managing-agents-challenges]]. Weigh the counter-view before over-committing: some governance experts favor **strict boundaries, audit trails, and human-retained liability** ('meaningful human control') over the 'AI subordinate' metaphor.


## Related across articles
- [[concept-agentic-workflows]]
- [[framework-agentic-report-generation]]