---
id: "framework-scenario-based-extraction"
type: "framework"
source_timestamps: ["§ Where to Start"]
tags: ["knowledge-extraction", "tactics", "tacit-knowledge"]
related: ["concept-codifying-judgment", "contrarian-experts-cannot-document", "action-convene-expert-panels"]
steps: ["Convene a small panel of experienced practitioners in the same role.", "Bring in a skilled moderator.", "Walk the group through a series of realistic scenarios and actual edge cases.", "Identify clear policies where the panel agrees quickly.", "Capture the debate where the panel disagrees to surface tacit judgment.", "Use the transcript of the conversation as the first draft of codified judgment for AI agents."]
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-new-27-teach-ai-your-decisions"
sourceUrl: "https://hbr.org/2026/06/teach-your-ai-how-you-make-decisions"
sourceTitle: "Teach Your AI How You Make Decisions"
---
# Scenario-Based Judgment Extraction

A tactical framework for extracting tacit knowledge from domain experts, bypassing the failure mode of asking them to document their processes abstractly (see [[contrarian-experts-cannot-document]]). Because experts know more than they can articulate directly, organizations should use *debate* to externalize reasoning (see [[quote-debate-externalizes-reasoning]]).

**Steps:**
1. Convene a small panel of experienced practitioners in the same role.
2. Bring in a skilled moderator.
3. Walk the group through a series of realistic scenarios and actual edge cases.
4. Identify clear policies where the panel agrees quickly.
5. Capture the debate where the panel *disagrees* — this is where tacit judgment surfaces.
6. Use the transcript as the first draft of codified judgment for AI agents.

The resulting transcript serves as the foundational context layer for AI agents, capturing nuance about risk tolerance, empathy, and escalation logic that standard documentation misses. It is operationalized by [[action-convene-expert-panels]] and [[action-use-transcripts-as-context]], feeding [[concept-codifying-judgment]].

**Enrichment note:** This mirrors established elicitation methods — the Critical Decision Method and cognitive task analysis used in emergency services and aviation, and the "critical incidents" technique in knowledge management — giving it credible methodological lineage even though its use as AI context is newly proposed.