---
id: "contrarian-decentralized-over-siloed-ai"
type: "contrarian-insight"
source_timestamps: ["§ Three Guiding Principles"]
tags: ["contrarian-insight", "organizational-design", "innovation"]
related: ["concept-decentralized-innovation-at-scale", "concept-generative-intelligence-group"]
challenges: "The traditional corporate instinct to centralize and silo emerging technology development within specialized R&D or IT departments."
sources: ["execution"]
sourceVaultSlug: "hbr-seg-execution"
originDay: 8
articleStem: "hbr-cl-93-legacy-financial-all-in-genai"
sourceUrl: "https://hbr.org/2025/03/how-a-legacy-financial-institution-went-all-in-on-gen-ai"
sourceTitle: "How a Legacy Financial Institution Went All In on Gen AI"
---
# Decentralized Access Over Specialized AI Divisions

## Contrarian Insight: Decentralized Access Over Specialized AI Divisions

**Challenges:** the traditional corporate instinct to centralize and silo emerging-technology development within specialized R&D or IT departments.

When adopting a major new technology, enterprises typically create a dedicated, centralized division (a 'Chief AI Office' or specialized R&D lab) to control development. [[entity-moodys|Moody's]] took the **opposite** approach — treating AI as a **bottom-up capability** and giving raw tools to all **14,000 employees**, using a central group only for **enablement and security** rather than primary development.

### Connections
- The concept: [[concept-decentralized-innovation-at-scale]].
- The enablement mechanism that makes it safe: [[concept-generative-intelligence-group]].

### Steelman / counter-perspective (from enrichment)
Decentralized experimentation can create **sprawl and governance overhead**. Handing GenAI to 14,000 people surfaces many use cases but also increases the burden on security, compliance, IT, and change-management functions. In regulated industries, bottom-up innovation often needs **stronger guardrails** than the 'everyone can innovate' framing suggests — which is precisely why the [[concept-generative-intelligence-group|GiG]] and embedded compliance ([[action-integrate-risk-and-compliance]]) are load-bearing, not optional.


## Related across articles
- [[contrarian-no-single-ai-hero]]
- [[claim-every-leader-a-shaper]]
- [[concept-ai-center-of-excellence]]