---
id: "concept-prompt-craftsmanship"
type: "concept"
source_timestamps: ["§ What Makes an Effective Agent Manager"]
tags: ["skills", "prompt-engineering", "agent-training"]
related: ["framework-agent-manager-capabilities", "action-pair-managers-engineers", "concept-agent-manager"]
definition: "The skill of designing and refining the natural language instructions that govern an AI agent's behavior, functioning as the machine equivalent of employee training."
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-sig-58-agent-managers"
sourceUrl: "https://hbr.org/2026/02/to-thrive-in-the-ai-era-companies-need-agent-managers"
sourceTitle: "To Thrive in the AI Era, Companies Need Agent Managers"
---
# Prompt Craftsmanship

## Prompt Craftsmanship

The ability to design and **iteratively refine** the language and logic that shape an AI agent's behavior. The authors deliberately frame this not merely as technical 'prompt engineering' but as the **machine equivalent of employee training**.

It involves translating complex business logic into simple, adaptive **natural-language instructions** that shape an agent's **intent, judgment, and tone**.

### Connected notes
- One of the six competencies in [[framework-agent-manager-capabilities]].
- Motivates a division of labor: the domain-expert [[concept-agent-manager]] shapes intent/tone in natural language while a technical AI engineer handles deterministic execution — see [[action-pair-managers-engineers]].

### Enrichment note
Broadly corroborated as the central skill. Rasa lists 'prompt design' + 'continuous improvement' as pillars of agent management; Beam.ai frames managers as deeply engaged in 'prompt refinement and workflow optimization'; practitioner accounts pair prompt engineering with process mapping and workflow optimization as *learned* skills — reinforcing that this is a trainable craft, not a credential.


## Related across articles
- [[action-codify-into-markdown]]
- [[concept-codifying-judgment]]