KB Schema Design

Use an outcome-first structure. Give ChatGPT the role, goal, approved source notes, constraints, and exact output you want back.

I am using Build The System for L&D systems work.

Prompt: KB Schema Design

Goal: Design a minimum viable schema, metadata, and acceptance criteria for a performance knowledge base.

Audience: L&D teams making knowledge findable, maintainable, and safe to reuse.

Source material I will provide:
- Knowledge domain
- Primary users
- Current storage
- Maintenance owner
- Quality bar

Inputs to use: [domain], [users], [current storage], [owner], [quality bar]

Working notes:
- domain: [add domain]
- users: [add users]
- current storage: [add current storage]
- owner: [add owner]
- quality bar: [add quality bar]

Rules:
- Use only approved source material.
- Do not invent policy, compliance, learner, employee, customer, financial, or business facts.
- Do not paste sensitive, confidential, employee, learner, customer, or proprietary data into an AI tool unless your organization has approved that tool for that data.
- Treat AI output as a draft that needs human review.

Return:
1. Document hierarchy
2. Minimum viable schema
3. Bouncer criteria
4. Metadata tags

Human review checklist:
- Does the output stay inside the source material?
- Are assumptions and missing information labeled?
- Is there a clear human owner for the final decision?
- Does the output need legal, compliance, privacy, accessibility, leader, requester, or SME review?

Use labeled sections so context, source material, task, constraints, and output format stay separate.

<context>
I am using Build The System for L&D systems work.
Prompt: KB Schema Design
Use case: Design a minimum viable schema, metadata, and acceptance criteria for a performance knowledge base.
Audience: L&D teams making knowledge findable, maintainable, and safe to reuse.
</context>

<source_material_to_prepare>
- Knowledge domain
- Primary users
- Current storage
- Maintenance owner
- Quality bar
</source_material_to_prepare>

<inputs>
- domain: [add domain]
- users: [add users]
- current storage: [add current storage]
- owner: [add owner]
- quality bar: [add quality bar]
</inputs>

<task>
Use the source material and inputs to produce a practical, review-ready output for this L&D workflow. Stay inside the source material. If a detail is missing, mark it as "Needs confirmation" instead of inventing it.
</task>

<constraints>
- Use only approved source material.
- Do not invent policy, compliance, learner, employee, customer, financial, or business facts.
- Do not paste sensitive, confidential, employee, learner, customer, or proprietary data into an AI tool unless your organization has approved that tool for that data.
- Treat AI output as a draft that needs human review.
</constraints>

<output_format>
1. Document hierarchy
2. Minimum viable schema
3. Bouncer criteria
4. Metadata tags
</output_format>

<human_review>
- Does the output stay inside the source material?
- Are assumptions and missing information labeled?
- Is there a clear human owner for the final decision?
- Does the output need legal, compliance, privacy, accessibility, leader, requester, or SME review?
</human_review>

Use goal, source, expectations, and output. For Obsidian context, use approved excerpts, Word summaries, OneDrive files, SharePoint pages, Teams context, or Outlook threads.

Goal: Help me complete KB Schema Design for an L&D workflow.

Source: Use only the selected Microsoft 365 document, meeting notes, email thread, spreadsheet, SharePoint file, or pasted notes I provide. Do not use unsupported assumptions.

Context:
- Use case: Design a minimum viable schema, metadata, and acceptance criteria for a performance knowledge base.
- Audience: L&D teams making knowledge findable, maintainable, and safe to reuse.
- Source material to have ready:
- Knowledge domain
- Primary users
- Current storage
- Maintenance owner
- Quality bar

Inputs:
- domain: [add domain]
- users: [add users]
- current storage: [add current storage]
- owner: [add owner]
- quality bar: [add quality bar]

Expectations:
- Use only approved source material.
- Do not invent policy, compliance, learner, employee, customer, financial, or business facts.
- Do not paste sensitive, confidential, employee, learner, customer, or proprietary data into an AI tool unless your organization has approved that tool for that data.
- Treat AI output as a draft that needs human review.

Output:
1. Document hierarchy
2. Minimum viable schema
3. Bouncer criteria
4. Metadata tags

Before finalizing, add a short "Needs human review" section covering:
- Does the output stay inside the source material?
- Are assumptions and missing information labeled?
- Is there a clear human owner for the final decision?
- Does the output need legal, compliance, privacy, accessibility, leader, requester, or SME review?

Use a task-first structure with clear input labels, one example pattern, rules, and a concise output format. For Obsidian context, use approved excerpts, Drive exports, Google Docs, or NotebookLM source sets.

Task: Complete KB Schema Design for an L&D workflow.

Use case:
Design a minimum viable schema, metadata, and acceptance criteria for a performance knowledge base.

Audience:
L&D teams making knowledge findable, maintainable, and safe to reuse.

Input labels to use:
- domain: [add domain]
- users: [add users]
- current storage: [add current storage]
- owner: [add owner]
- quality bar: [add quality bar]

Source material to prepare:
- Knowledge domain
- Primary users
- Current storage
- Maintenance owner
- Quality bar

Example pattern:
Field: Missing information
Good answer: Name the specific detail to confirm, who can confirm it, and why it affects the next decision.

Rules:
- Use only approved source material.
- Do not invent policy, compliance, learner, employee, customer, financial, or business facts.
- Do not paste sensitive, confidential, employee, learner, customer, or proprietary data into an AI tool unless your organization has approved that tool for that data.
- Treat AI output as a draft that needs human review.

Return:
1. Document hierarchy
2. Minimum viable schema
3. Bouncer criteria
4. Metadata tags

Human review checklist:
- Does the output stay inside the source material?
- Are assumptions and missing information labeled?
- Is there a clear human owner for the final decision?
- Does the output need legal, compliance, privacy, accessibility, leader, requester, or SME review?