---
id: "framework-gen-ai-deployment"
type: "framework"
source_timestamps: ["§ Where and When to Use Generative AI", "§ Applying the Framework"]
tags: ["decision-matrix", "task-allocation", "strategy"]
related: ["concept-cost-of-errors", "concept-knowledge-type-tacit-vs-explicit", "concept-no-regrets-zone", "concept-creative-catalyst-zone", "concept-human-first-zone", "concept-quality-control-zone", "action-deconstruct-jobs"]
steps: ["Break down existing jobs into their component tasks.", "\\\"Assess the 'Cost of Errors' for each task (would a mistake cause serious financial", "legal", "or reputational harm?).\\\"", "\\\"Assess the 'Type of Knowledge' required for each task (does it rely on explicit", "structured data", "or tacit", "nuanced human judgment?).\\\"", "\\\"Plot the task into one of the four quadrants (No Regrets", "Creative Catalyst", "Quality Control", "Human-First).\\\"", "Deploy Gen AI according to the rules of that specific quadrant (autonomous deployment vs. human-in-the-loop vs. strictly supportive)."]
source_url: "https://hbr.org/2025/11/the-gen-ai-playbook-for-organizations"
source_title: "The Gen AI Playbook for Organizations"
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-cl-87-genai-playbook-orgs"
sourceUrl: "https://hbr.org/2025/11/the-gen-ai-playbook-for-organizations"
sourceTitle: "The Gen AI Playbook for Organizations"
---
# Framework for Choosing Where and How to Use Gen AI

The **central contribution of the article**: a **2×2 matrix** for identifying the best places to deploy generative AI by breaking jobs down into component tasks. The framework plots each task along two dimensions — the [[concept-cost-of-errors|Cost of Errors]] (Low vs. High) and the [[concept-knowledge-type-tacit-vs-explicit|Type of Knowledge]] required (Explicit vs. Tacit) — producing four zones:

1. **[[concept-no-regrets-zone|No Regrets Zone]]** (Low cost of error · Explicit knowledge): Deploy AI immediately for speed and scale — e.g., resume screening, meeting summaries. Where AI agents will thrive.
2. **[[concept-creative-catalyst-zone|Creative Catalyst Zone]]** (Low cost of error · Tacit knowledge): Use AI to augment creativity and generate volume/variations — e.g., marketing taglines, design mock-ups.
3. **[[concept-quality-control-zone|Quality Control Zone]]** (High cost of error · Explicit knowledge): Use AI for heavy lifting but mandate a human-in-the-loop for accountability — e.g., legal contracts ([[entity-harvey|Harvey]]), software code ([[entity-github-copilot-d6|GitHub Copilot]]).
4. **[[concept-human-first-zone|Human-First Zone]]** (High cost of error · Tacit knowledge): Constrain AI to a supportive role; humans remain the central decision-makers — e.g., executive hiring, strategy setting.

**How to apply it (steps):**
1. Break existing jobs into component tasks (see [[action-deconstruct-jobs]]).
2. Assess the *cost of errors* for each task.
3. Assess the *type of knowledge* each task requires.
4. Plot the task into one of the four quadrants.
5. Deploy according to that quadrant's rules (autonomous vs. human-in-the-loop vs. strictly supportive).

**Why it matters:** It reframes the question from *"is the AI smart enough?"* to *"where can we safely capture value now?"* and it operationalizes the article's answer to the replacement-vs-complementarity debate (see [[quote-replacement-vs-complementarity]]). The framework is directly validated by institutional summaries (NYU Stern, HBS) and is grounded in Polanyi's epistemology and risk-based deployment thinking (see [[prereq-tacit-vs-explicit-knowledge-d6]]). *Refinement to keep in mind:* many real tasks blend tacit and explicit elements, so treat quadrant boundaries as fuzzy.