MMarketing Against The Grain
← All frameworks
Productivity

Guide-to-Prompt Library

Convert an authoritative prompt guide into task-ready templates for your field

Difficulty
Easy
Time to result
~days to results
Steps
5
Confidence
96%

The Guide-to-Prompt Library converts an authoritative document into a structured, domain-specific reference. The user supplies a prompt engineering guide and instructs an AI to identify the prompt types it explicitly teaches, without inventing additional categories. For every type, the AI produces a detailed example, explains the conditions under which it works best, and translates it into a relevant professional use case. A marketer, for example, receives marketing prompts rather than abstract demonstrations alone. The resulting table reduces the distance between learning a technique and applying it. Human review remains important: reading the original guide provides context, while checking the generated library protects against omissions, invented classifications, and shallow examples.

Origin

Extracted from Marketing Against the Grain, where a Google prompt engineering white paper was converted into a practical prompt library for marketers.

Core principles

  • 01Use authoritative source material as the boundary
  • 02Preserve the source guide's prompt taxonomy
  • 03Explain when each prompt type should be used
  • 04Translate abstract techniques into domain-specific examples
  • 05Keep human understanding alongside AI-assisted extraction

How to run it

  1. 1

    Choose the source guide

    Select a credible prompt engineering guide that contains enough concrete techniques to support a reusable library. Use the full document rather than an isolated excerpt when possible.

    Pro tip Multiple reputable guides can later be combined to broaden coverage.

    Watch out A weak or promotional source will produce a weak library.

  2. 2

    Constrain the extraction

    Tell the AI to identify only the prompt types taught in the guide and to retain the source's distinctions. Explicitly prohibit invented prompt categories.

    Pro tip Request a table so every prompt type follows the same structure.

    Watch out Without source constraints, the model may add familiar techniques that are absent from the guide.

  3. 3

    Explain appropriate use

    For each extracted prompt type, specify when it should be selected and what kind of task or uncertainty it handles. This turns a list of names into a practical decision resource.

    Pro tip Describe both the positive use case and situations where another prompt type would work better.

    Watch out Definitions without selection guidance are difficult to apply during real work.

  4. 4

    Generate detailed examples

    Create one clear general example for each technique, followed by an example adapted to the target profession or domain.

    Pro tip Include realistic context, constraints, audience, and desired output in domain examples.

    Watch out Short toy examples can hide how much context a real task requires.

  5. 5

    Validate against the guide

    Compare the generated library with the source document to confirm that prompt types, descriptions, and examples are faithful. Refine any entries that are vague or unsupported.

    Pro tip Read the source before finalizing the library so you understand what the AI extracted.

    Watch out Skipping review trades genuine understanding for unverified automation.

In the wild

Marketing prompt reference

A marketer uploads Google's prompt engineering guide and asks an AI to extract each documented prompt type. The resulting table explains zero-shot, few-shot, persona, and contextual prompting, then supplies marketing examples such as writing launch-email subject lines for a luxury skincare audience.

The marketer receives a role-specific prompt reference that can be used immediately during campaign work.

Customer-support playbook

A support leader applies the same workflow to a reputable AI prompting guide. Each extracted technique receives a support-specific example covering ticket classification, tone matching, escalation summaries, and policy-grounded answers.

The support team gains a consistent library for selecting and writing prompts without inventing a new methodology.

Common mistakes

Inventing unsupported prompt types

Allowing the AI to supplement the guide weakens source fidelity and makes it unclear which techniques came from the authoritative document.

Using generic examples only

Abstract demonstrations may teach syntax but fail to show how the technique operates under realistic professional constraints.

Skipping the source document

Delegating all reading to AI may save time, but it leaves the user unable to judge whether the resulting library is accurate or complete.

Is it for you?

Best for

Professionals who need to translate authoritative technical guidance into reusable prompts for a specific role or industry.

Not ideal for

Users who lack a trustworthy source guide or need wholly original prompt techniques beyond the source material.

From the transcript

The very first thing we do is we create a prompt to turn that PDF into a range of templates that we can apply to…

01:30

Tell me when I should actually use that prompt and then give me a marketing specific example of how to use that prompt.

02:00

don't invent new prompt types. Only use the things from the guide.

02:30

From the episode

Become a Google-Level Prompt Engineer in 20 Minutes