---
id: "action-analyze-task-level"
type: "action-item"
source_timestamps: ["§ Three Necessities", "¶14", "¶15"]
tags: ["workforce-analytics", "ai-integration"]
related: ["concept-continuous-sensing", "entity-stripe-minions", "entity-github-copilot"]
speakers: ["Sangeet Paul Choudary", "John Winsor"]
action: "Track task-level data to see exactly which workflow components are absorbed by AI versus handled by humans."
outcome: "Clear visibility into the shifting division of labor, preventing the organization from hiring for obsolete skills."
sources: ["tail1"]
sourceVaultSlug: "hbr-seg-tail1"
originDay: 1
articleStem: "hbr-tail-112-continually-assessing-performance"
sourceUrl: "https://hbr.org/2026/06/the-pros-and-cons-of-continually-assessing-performance"
sourceTitle: "The Pros and Cons of Continually Assessing Performance"
---
# Analyze Work at the Individual Task Level

**Action:** Track task-level data to see exactly which workflow components are absorbed by AI versus handled by humans.

**Do this:** Implement [[concept-continuous-sensing]] to track which parts of a task are delegated to AI, which are rewritten by humans, and which pass final review. Evidence patterns: [[entity-stripe-minions]] (1,300+ AI-written, human-reviewed submissions merged/week) and [[entity-github-copilot-d1]] telemetry (33% suggestion-acceptance, 20% line-acceptance across 400+ developers, per [[entity-zoominfo]]).

**Expected outcome:** Clear visibility into the shifting division of labor, preventing the organization from hiring or organizing for obsolete skills.

This is Necessity #2 of the [[framework-three-necessities]]. Read the signals through the lens of [[contrarian-productivity-vs-capability]] — reward supervision of AI, not just accelerated output.