---
id: "claim-agent-insertion-fails"
type: "claim"
source_timestamps: ["§ Three Ways to Respond"]
tags: ["process-engineering", "deployment-strategy"]
related: ["framework-three-responses", "concept-documented-organization"]
confidence: "high"
testable: true
speakers: ["K. Sudhir"]
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-new-26-agentic-systems-implicit-rules"
sourceUrl: "https://hbr.org/2026/06/how-to-design-agentic-systems-around-the-implicit-rules-that-govern-your-company"
sourceTitle: "How to Design Agentic Systems Around the Implicit Rules that Govern Your Company"
---
# Agent insertion and naive reengineering fail

**Claim (confidence: high · testable):** Two of the three common deployment paths fail:

- **Agent insertion** — dropping agents into existing workflows — fails because the [[concept-implicit-organization]] does not adapt, leaving the agent to run a brittle, partial version of the process.
- **Naive reengineering** — redesigning around agent capabilities based only on the [[concept-documented-organization]] — also fails because it merely *recreates the organization's blind spots at machine speed.*

Only the third path, **informed reengineering**, succeeds. See the full taxonomy in [[framework-three-responses]].

**Enrichment / confidence calibration:** Strong conceptual support — aligning AI systems with actual work practices (including tacit elements) is consistent with information-systems and organizational-change research (socio-technical alignment; ERP/workflow failures when built on formal process alone). The categorical 'fail' should be read as 'often fail' / 'are prone to failure' rather than universally failing.