Questionnaire
~2 min fill
→
9 Prompt Libraries
Run in parallel
→
Fan-In Engine
9 outputs in
→
Knowledge Base
< 4 minutes
Prompt Library ArchitecturePrompt Library Deep Dive Series

How 23 Column Prompts

One questionnaire. Twenty-three prompts. A complete company identity — every value proposition, every mission statement, the full brand package — generated in under four minutes.

T
Tommy Saunders
Founder, The Prompt Engineering Project
March 15, 2026
\u00b7 9 min read
PEP-PL-2026-001SERIES PLAN
Prompt Sources
Company IdentitySEO & Web CopyCompany IdentityCompany IdentityCompany IdentityCompany IdentityCompany IdentitySocial Media
\u25b8Direct Answer

What is the Prompt Library System and how do 23 column prompts work?

The Prompt Library System dispatches one questionnaire to nine specialized prompt libraries containing 23 column prompts — Company Identity, Content Strategy, Target Audience, Social Media, Brand Identity, SEO & Web Copy, Sales Enablement, Email Marketing, and Editorial Standards — producing a complete, coherent knowledge base in under four minutes. Each prompt column executes through a fan-out architecture and returns structured outputs to a central Fan-Out to Fan-In Engine, which assembles the final knowledge base without accumulating context overhead. Strategic coherence across all outputs is guaranteed by architecture: every prompt library reads the same questionnaire, not each other’s outputs.

Source: thepromptengineeringproject.com · Prompt Library System · March 2026JSON-LD Schema

Here is what most prompt engineering workflows get wrong. They write the value proposition prompt. Then the audience prompt. Then the brand voice prompt. Then the SEO prompt. Each prompt works from a different version of the same brief, on a different timeline, with a different interpretation of what the brand should sound like.

The result is five pieces of output that don’t quite match, published across five different channels with five different voices. The company identity says disruption. The social post promises transformation. The email says something else entirely. The message is fractured before it ever reaches the customer.

The Prompt Library System runs all nine disciplines from one questionnaire — simultaneously. Company Identity prompt library. Content Strategy prompt library. Target Audience prompt library. Social Media distribution. Brand Identity system. SEO & Web Copy. Sales Enablement sequence. Email Marketing. Editorial Standards curation. Every discipline fires in parallel from the same questionnaire, and every output feeds into a single assembled knowledge base.

“One questionnaire. Nine prompt libraries running in parallel. The entire knowledge base — synchronized by architecture, not by editorial review.”

Tommy Saunders · Founder, The Prompt Engineering Project

Social Distribution Suite

Social Distribution SuitePrompt Library Deep Dive Series
T
Tommy Saunders
@tommysaunders_ai
Nine prompt libraries. One questionnaire. That’s the Prompt Library System. Company identity + content strategy + social + SEO + sales enablement — all from a single questionnaire input, all running simultaneously. Here’s the architecture behind it
3:00 PM · Mar 15, 2026 · 14.2K Impressions
Search Package — PEP-PL-2026-001
thepromptengineeringproject.com › products › prompt-libraries
How 23 Column Prompts Build One Knowledge Base | The Prompt Engineering Project
See how The Prompt Engineering Project runs 9 specialized prompt libraries from one questionnaire — producing a complete company identity, content strategy, social posts, and sales enablement sequence simultaneously in under 4 minutes.
prompt library systemcolumn promptsprompt engineering workflowquestionnaire AIknowledge base generationNotion AI templatesprompt-chain architecturestructured knowledge base
5-Step Nurture Sequence — PEP-PL-2026-001 CRM Output
Day 0Architecture guide + article link delivered
Day 2How the questionnaire actually works
Day 5Case study: Prompt library in action
Day 8Demo invitation: run your questionnaire live
Day 14Still thinking about it?

Frequently Asked Questions

5 Questions
The Prompt Library System dispatches one questionnaire to nine specialized prompt libraries containing 23 column prompts — Company Identity, Content Strategy, Target Audience, Social Media, Brand Identity, SEO & Web Copy, Sales Enablement, Email Marketing, and Editorial Standards. Each prompt library executes through its own prompt chain, produces its deliverables, and returns a structured output to the Fan-Out to Fan-In Engine. The engine assembles all nine outputs into a complete, publication-ready knowledge base in under four minutes. Strategic coherence across outputs is guaranteed by architecture: every prompt library reads the same questionnaire, never each other’s outputs.
Legacy prompt tools accumulate every working note, failed attempt, and internal reasoning step in their context window. By step 20-30, the context window is so full of noise that performance degrades — what engineers call the “Dumb Zone.” The Prompt Library System’s prompt-chain architecture compresses each library’s output into a structured JSON object. The Fan-Out to Fan-In Engine reads only clean state deltas, keeping its context window virtually flat regardless of how many steps have executed. This enables stable performance past 1,000 steps where legacy architectures fail.
A Questionnaire is the single structured input document that every prompt library reads from. It takes approximately two minutes to complete and contains: brand identity, industry category, target audience (primary and secondary), key message, competitive context, visual style direction, and SEO cluster. Every prompt library reads from this one source simultaneously — which is what guarantees strategic alignment by architecture rather than editorial review. The questionnaire never changes between pipeline runs for the same campaign.
Tools like Jasper and Copy.ai are single-prompt tools — you prompt them once and receive one type of output. The Prompt Library System runs nine specialized prompt libraries in parallel from a single questionnaire, with each library optimized for its specific discipline. More importantly, the architectural difference is coherence: because every prompt library reads the same questionnaire, the LinkedIn post reinforces the company identity’s argument rather than contradicting it. This cross-channel coherence is extremely difficult to achieve with single-prompt tools regardless of how much editorial coordination is applied.
Fan-out dispatch means all nine prompt libraries begin executing simultaneously the moment the questionnaire is submitted — rather than waiting for each library to complete before starting the next. In practice, the Company Identity Library begins generating value propositions while the Content Strategy Library builds topic clusters, the Target Audience Library outlines persona profiles, and the SEO & Web Copy Library builds keyword clusters — all at the same time. This is why the complete knowledge base arrives in under four minutes despite nine disciplines executing. None of them wait for the others.
Structured as FAQ schema (JSON-LD) for AEO indexing

References

1The structured knowledge base architecture is documented in the Prompt Library System technical brief: “From Questionnaire to Knowledge Base — How Prompt-Chain Architectures Maintain Reasoning Quality at Scale,” thepromptengineeringproject.com, 2026. The core insight derives from OS engineering: treating the LLM as a CPU and the context window as RAM, and managing each with the discipline of a systems programmer rather than a chatbot developer.
2The distinction between “one questionnaire, nine outputs” versus “nine questionnaires, nine outputs” is not merely operational efficiency — it is a structural guarantee of coherence. When the Social Media Library reads the same competitive context as the Company Identity Library, the LinkedIn post reinforces the company identity’s argument rather than contradicting it. This coherence is extremely difficult to achieve in sequential workflows regardless of how much editorial coordination is applied.
← Back to series