---
id: "concept-agentic-ai-d7"
type: "concept"
source_timestamps: ["¶2"]
tags: ["ai-architecture", "autonomy"]
related: ["concept-personal-ai-agents", "claim-micromanagement-defeats-purpose", "prereq-principal-agent-problem", "quote-agentic-ai-definition", "entity-salesforce"]
definition: "A programming layer on top of LLMs that enables autonomous goal-seeking behavior, including data collection, decision-making, system interaction, and behavioral adaptation."
sources: ["governance"]
sourceVaultSlug: "hbr-seg-governance"
originDay: 7
articleStem: "hbr-cl-88-can-ai-agents-be-trusted"
sourceUrl: "https://hbr.org/2025/05/can-ai-agents-be-trusted"
sourceTitle: "Can AI Agents Be Trusted?"
---
# Agentic AI

Agentic AI represents a distinct architectural layer built on top of foundational Large Language Models (LLMs). Unlike standard LLMs that passively generate text in response to user prompts, agentic AI systems are designed to work autonomously toward specific, user-defined goals. This software layer possesses the capability to collect data, make independent decisions, take concrete actions, interact with external systems, apply reasoning, and adapt its behavior based on the results of its actions. By operating according to priorities and rules set by the user (the principal), agentic AI shifts the paradigm from human-computer interaction to human-computer delegation.

This delegation is precisely what creates the [[prereq-principal-agent-problem]] at the heart of the source's thesis. The consumer-facing application of this layer is the [[concept-personal-ai-agents]] discussed throughout the article. Because the value of an agent lies in acting *without* constant oversight, any attempt to secure it through supervision runs headlong into [[claim-micromanagement-defeats-purpose]]. See the authors' own working definition in [[quote-agentic-ai-definition]]. Early *enterprise* deployment (distinct from the riskier consumer frontier) is exemplified by [[entity-salesforce-d7]].


## Related across articles
- [[concept-agentic-governance]]
- [[concept-agentic-ai-governance-gap]]
- [[prereq-agentic-ai-concepts]]
- [[prereq-agentic-ai-understanding]]