---
id: "concept-learning-in-the-flow-of-work"
type: "concept"
source_timestamps: ["§ They Train People in the Context of Real Work", "¶12", "¶13"]
tags: ["training", "contextual-learning", "real-time-analytics"]
related: ["concept-co-learning", "claim-traditional-training-metrics-fail", "prereq-real-time-data-infrastructure", "action-shift-to-in-flow-training", "framework-building-ai-with-workers"]
definition: "Context-specific, real-time training delivered directly on the factory floor during actual operations, rather than in isolated classrooms."
sourceUrl: "https://hbr.org/2026/05/the-best-manufacturers-build-ai-with-workers-not-for-them"
sourceTitle: "The Best Manufacturers Build AI with Workers, Not for Them"
sources: ["adoption"]
sourceVaultSlug: "hbr-seg-adoption"
originDay: 9
articleStem: "hbr-cl-78-build-ai-with-workers"
---
# Learning in the Flow of Work

Most traditional training programs remove workers from their environments to study broad concepts in isolation. **Learning in the flow of work** is the antithesis of this approach: it provides detailed, context-specific training directly on the factory floor *as work is being performed*.

If operators are expected to use AI tools to diagnose production issues, adjacent training that only covers those tasks in broad strokes is ineffective. Workers need to learn alongside the reoptimizing systems and algorithmic recommendations they are expected to validate. Supported by real-time analytics, this contextual learning lets supervisors see exactly where work stalls, where errors rise, or where worker confidence drops — so they can intervene and coach effectively.

**Worked example.** At a food-processing plant, operators learned to use a *next-best-action* dashboard in real time on the line, adjusting line speed and seasoning equipment *before* waste occurred. Because the training was immediately relevant and situated in their actual environment, yield improved and system acceptance rose rapidly.

This is Pillar 2 of the [[framework-building-ai-with-workers]]. It is operationalized by [[action-shift-to-in-flow-training]], hard-depends on [[prereq-real-time-data-infrastructure]] (without live data you cannot see where work stalls), and is the setting in which [[concept-co-learning]] occurs. It is the practical alternative to the participation metrics debunked in [[claim-traditional-training-metrics-fail]].

> **Counter-perspective to hold in mind.** In-flow training is not a blanket replacement for classroom instruction. Off-line training remains essential for foundational concepts, safety certification, regulated procedures, and low-frequency/high-risk scenarios. The enrichment supports continuous learning but does not imply that *all* formal instruction should be abandoned.