AI Workflow Operator Course for Beginners

Step-by-step workflow

• Human: Choose one repeated task from real work.
• Human: Gather the source material, such as notes, transcript, draft, or report.
• Human: Remove private or unnecessary information.
• AI: Sort the task into human decisions, AI-helpful steps, and review points.
• Human: Check whether the AI misunderstood the task or invented details.
• Human: Save the useful parts as the start of a workflow canvas.

Copy-paste prompt

I want to turn one repeated work task into a safe AI workflow.

Task:
[Describe the task]

Current process:
[List the steps you usually take]

Source material I usually have:
[Notes, transcript, spreadsheet, brief, email thread, customer questions, etc.]

Please separate this task into:
1. Decisions a human should still make
2. Steps AI could help with, such as drafting, summarizing, sorting, comparing, or formatting
3. Information the AI would need before starting
4. Output I should ask AI to produce
5. Human review checkpoints before the output is used
6. Risks or warning signs for a beginner

Rules:
- Keep the workflow beginner-friendly.
- Do not suggest full automation yet.
- Do not invent facts about my work.
- If something is unclear, mark it as "Needs clarification."

Expected output

A good response should give you a clear split between human judgment and AI assistance. It should name the input needed, suggest a simple output format, flag unclear parts, and identify review points before anything is sent, published, changed, or decided.

What can go wrong

• The AI suggests automating more than you can safely review.
• The task is too broad, so the workflow becomes vague.
• The AI misses a decision that should stay human-owned.
• The output sounds organized but does not match how the work really happens.

How to improve the workflow

Narrow the task, add a better example of the source material, define the output format more clearly, and add a review checklist. If the workflow still feels hard to check, choose a simpler first task.

Mini exercise

Choose one task you repeat at least once a week. Write three columns: human decisions, AI-helpful steps, and review checkpoints. Then run the prompt above and compare the AI’s breakdown to your own.

Step-by-step workflow

• Human: Describe the task you are considering.
• Human: Identify the input, output, audience, and consequence if wrong.
• AI: Rate the task for repeatability, clarity, reviewability, risk, and source quality.
• Human: Review the rating and adjust it using your real-world context.
• Human: Decide whether to use AI now, use AI for only part of the task, or avoid the task as a beginner workflow.

Copy-paste prompt

Help me decide whether this task is a good fit for beginner AI assistance.

Task:
[Describe the task]

Where the output will be used:
[Internal notes, customer email draft, public content, report, spreadsheet, CRM, etc.]

Input material available:
[Notes, transcript, document, spreadsheet, source links, customer questions, or none]

Please rate the task from 1 to 5 on:
- Repeatability
- Clarity of input
- Clarity of output
- Ease of human review
- Risk if wrong
- Need for current or verified facts
- Sensitivity of data

Then recommend one of these:
1. Good beginner AI workflow
2. Use AI for a small internal part only
3. Do not automate this as a beginner

Explain your reasoning in plain English.
Also list the human approval checkpoint that should happen before the output is used.

Expected output

A good response should give a practical rating, not a hype answer. It should explain the task’s strengths and risks, name the safest AI-assisted part, and recommend a clear human review checkpoint. If the task is high-risk, it should say so plainly.

What can go wrong

• AI recommends a task that is too risky because it does not understand the workplace consequences.
• AI treats missing source material as permission to guess.
• A polished draft includes unsupported public claims.
• A workflow touches customer data without a privacy review.
• The human review step is vague, so no one knows what to check.

How to improve the workflow

Add clearer task boundaries, provide real source material, define who reviews the output, and remove any action that sends, publishes, deletes, buys, updates records, or makes decisions automatically.

Mini exercise

Pick three tasks from your week: one content or writing task, one operations or admin task, and one customer or marketing task. Run the prompt for each. Put them into three buckets: good first workflow, AI-assisted draft only, and not a beginner automation candidate.

Step-by-step workflow

• Human: Choose one prompt you have used before or one task you keep doing manually.
• Human: Define the trigger for when the workflow should run.
• Human: List the input material the AI needs.
• AI: Perform one narrow task and produce a structured output.
• Human: Review the output against a checklist.
• Human: Approve, revise, save, send, publish, or stop based on the review.

Copy-paste prompt

Turn this one-off AI prompt into a repeatable workflow.

Prompt I currently use:
[Paste your prompt]

The work goal:
[Explain what this prompt helps you do]

Where the output will be used:
[Internal notes, public content, customer reply draft, report, SOP, etc.]

Please create:
1. Trigger: when to use this workflow
2. Input checklist: what I need before using AI
3. AI task: the narrow job AI should perform
4. Output format: the sections or table the AI should produce
5. Human review checklist: what I must check before use
6. Next action: what happens after review
7. Stop signs: when I should not use this workflow

Rules:
- Keep the workflow beginner-friendly.
- Do not suggest full automation yet.
- Include a human approval point.
- Mark any risky or unclear part as "Needs review."

Expected output

A good output should give you a simple workflow map. It should be clear enough that you can run it again next week without rewriting everything. It should also name the review point and any stop signs.

What can go wrong

• The workflow is still too broad to repeat reliably.
• The trigger is unclear, so people use the workflow inconsistently.
• The input is weak, so the AI fills gaps with guesses.
• The output format is vague, so review takes too long.
• The next action skips human approval.

How to improve the workflow

Narrow the AI task, improve the input checklist, add examples of good output, strengthen the review checklist, and create stop signs for cases that are too risky or unclear.

Mini exercise

Find one prompt you have used before. Turn it into a five-part workflow: trigger, input, AI task, review, and next action. Then ask yourself whether someone else could follow it without you explaining it live.

Filled example: meeting summary workflow

Filled example: YouTube video research workflow

Filled example: sponsor lead research workflow

Blank AI Workflow Canvas students can copy

AI Workflow Canvas

Task:
[What repeated task will this workflow help with?]

Trigger:
[When should this workflow be used?]

Input:
[What source material does the AI need?]

Tool:
[What AI tool or work tool will be used?]

Prompt:
[What reusable instruction will be pasted into the AI tool?]

Output:
[What should the AI produce?]

Human review point:
[Where must a human stop and check the output?]

Risk level:
[Low, medium, or high, and why?]

Next action:
[What happens after review?]

Success standard:
[How will you know the workflow worked?]

Failure warning signs:
[What shows the workflow is not reliable yet?]

Privacy notes:
[What information should be removed, masked, or handled carefully?]

Reuse frequency:
[How often will this workflow be used?]

Owner:
[Who is responsible for running, reviewing, and improving this workflow?]

Copy-paste prompt

Help me fill out an AI Workflow Canvas for this task.

Task:
[Describe the repeated task]

Current process:
[Briefly describe how I do it now]

Where the output will be used:
[Internal, customer-facing draft, public content, report, SOP, etc.]

Source material available:
[Notes, transcript, document, spreadsheet, approved sources, etc.]

Please create a beginner-friendly AI Workflow Canvas with these fields:
- Task
- Trigger
- Input
- Tool
- Prompt
- Output
- Human review point
- Risk level
- Next action
- Success standard
- Failure warning signs
- Privacy notes
- Reuse frequency
- Owner

Rules:
- Keep the first version narrow and reviewable.
- Do not suggest full automation yet.
- Mark unclear or risky parts as "Needs review."
- Include a human approval point before anything is sent, published, changed, or decided.

Expected output

A good output should give you a complete but simple canvas. It should not be a giant system design. The task should be narrow, the input should be clear, the prompt should be reusable, and the review point should happen before any important action.

What can go wrong

• The task is too broad, so the canvas becomes vague.
• The input is not defined, so AI guesses.
• The output is not structured, so review takes too long.
• The risk level is understated.
• The owner is missing, so no one improves the workflow.

How to improve the workflow

After testing, update the canvas. Add clearer input requirements, tighten the prompt, improve the output format, add stronger failure warning signs, and make the review checklist more specific.

Mini exercise

Choose one task from your work and fill in the blank canvas. Then compare it to the three filled examples. If your task has no clear trigger, input, review point, or owner, simplify it before using AI.

Step-by-step prompt repair workflow

• Human: Paste the weak prompt and explain the real work goal.
• Human: Add context: audience, source material, output use, and constraints.
• Human: Define success criteria and review criteria.
• AI: Rewrite the prompt as a reusable workflow prompt.
• AI: Include assumptions, caveats, uncertainty, what it used, and what it ignored.
• Human: Test the prompt on one realistic example before reusing it.

Copy-paste prompt

Improve this prompt so it works inside a repeatable AI workflow.

Weak prompt:
[Paste the prompt]

Workflow goal:
[Explain the real work goal]

Audience or user of the output:
[Who will use or read the output?]

Source material available:
[Notes, transcript, document, spreadsheet, approved sources, etc.]

How the output will be used:
[Internal draft, customer-facing draft, public content, planning document, SOP, etc.]

Please rewrite the prompt with:
1. Role
2. Goal
3. Context
4. Source material instructions
5. Output format
6. Success criteria
7. Rules and constraints
8. Questions to ask if information is missing
9. Assumptions to list
10. Caveats to list
11. Uncertainty to flag
12. A section showing what source material was used and what was ignored
13. A human review reminder

Rules:
- Do not make the prompt longer than needed.
- Do not invent facts.
- Keep it practical for a beginner.
- Include a clear human approval point before the output is used.

Expected output

The output should be a reusable prompt template, not a one-time answer. It should include placeholders for source material, a clear output format, success criteria, uncertainty labels, assumptions, caveats, and a human review reminder.

What can go wrong

• The prompt becomes long but still does not define success.
• The prompt asks for caveats, but no one reviews them.
• The AI lists assumptions and the human treats them as facts.
• The prompt asks for source use, but the source material is weak.
• The prompt is reused for a riskier task than it was designed for.

How to improve the workflow

After the first run, save what the AI misunderstood. Add better source material, tighten the output format, add examples of good output, and strengthen the review checklist. Remove any instruction that encourages guessing.

Mini exercise

Take one prompt you already use. Add context, success criteria, assumptions, caveats, uncertainty, and a “what you used / what you ignored” section. Test it with one realistic input and mark what still needs human review.

Beginner workflow 1: Meeting summary

When to use it

Use this after a client call, internal project meeting, podcast planning call, sales conversation, or team check-in.

Inputs needed

Meeting notes, transcript excerpt, attendee names if relevant, known deadlines, and project context.

Tool suggestion

AI assistant plus a document editor. Keep the final version in the place your team already uses.

Copy-paste prompt

You are helping me turn messy meeting notes into a reviewable follow-up summary.

Meeting context:
[Briefly describe the meeting]

Meeting notes:
[Paste notes or transcript excerpt]

Please produce:
1. Short summary
2. Decisions made
3. Action items with owner only if clearly mentioned
4. Deadlines only if clearly mentioned
5. Open questions
6. Missing details or uncertainty
7. Draft follow-up message for human review

Rules:
- Do not invent owners, deadlines, decisions, or commitments.
- Mark unclear items as "Needs confirmation."
- Keep the tone clear and professional.
- Do not send anything or assume approval.

Expected output

A clean summary, action list, open questions, missing details, and a draft follow-up that is clearly marked for review.

Human review point

Before sending the follow-up or assigning tasks, confirm every owner, deadline, decision, and commitment.

Risk level

Low for internal summaries. Medium if the output will be sent to a client or customer.

How to improve it after the first run

Add a better meeting context, require a stricter action-item format, and add examples of how your team writes follow-ups.

Beginner workflow 2: Research brief

When to use it

Use this when you have notes, excerpts, links, or transcripts and need a structured internal brief before deciding what to do next.

Inputs needed

Source excerpts, notes, links for human reference, the research question, and what decision or content this research supports.

Tool suggestion

AI assistant plus a research document or spreadsheet. Use source material you can check.

Copy-paste prompt

You are helping me organize research from provided source material only.

Research question:
[What am I trying to understand?]

Source material:
[Paste notes, excerpts, transcript sections, or source summaries]

Please produce:
1. One-paragraph answer to the research question
2. Summary of each source
3. Key themes
4. Conflicts or differences between sources
5. Claims that need verification
6. Open questions
7. Suggested next research steps

Rules:
- Use only the source material I provided.
- Do not invent facts or sources.
- Label unsupported items as "Not confirmed."
- Keep the brief practical and easy to review.

Expected output

A source-based brief with summaries, themes, conflicts, verification needs, and next research questions.

Human review point

Before using the brief for strategy, content, or public claims, check the original sources.

Risk level

Low for internal learning. Medium if the research informs public content or business decisions.

How to improve it after the first run

Add clearer source labels, ask for a table format, and include a stronger rule against unsupported claims.

Beginner workflow 3: Content outline

When to use it

Use this before writing a blog post, newsletter, YouTube script, podcast episode, or landing page section from existing notes.

Inputs needed

Topic, audience, source notes, goal of the content, voice notes, and any claims that must be included or avoided.

Tool suggestion

AI assistant plus a writing document. Keep the final writing human-edited.

Copy-paste prompt

You are helping me turn source notes into a content outline for human review.

Content type:
[Blog post, newsletter, YouTube video, podcast, landing page, etc.]

Audience:
[Who is this for?]

Goal:
[What should the content help the audience understand or do?]

Source notes:
[Paste notes, transcript, outline, or approved talking points]

Please produce:
1. Working title options
2. Main promise or angle
3. Section-by-section outline
4. Key points for each section
5. Claims or examples that need verification
6. Gaps or questions before drafting

Rules:
- Keep the outline grounded in the source notes.
- Do not invent statistics, testimonials, or results.
- Avoid generic AI hype language.
- Mark uncertain items as "Needs confirmation."

Expected output

A clear outline with section headings, key points, gaps, and verification notes.

Human review point

Before drafting or publishing, check the angle, claims, voice, examples, and whether the outline is actually useful for the audience.

Risk level

Low for internal outlining. Medium if it will become public content.

How to improve it after the first run

Add examples of your preferred structure, define the audience more clearly, and remove sections that feel generic.

Beginner workflow 4: Competitor scan

When to use it

Use this when reviewing competitors, alternatives, sponsor candidates, or similar products from notes you collected manually.

Inputs needed

Public notes, page excerpts, pricing notes if verified, feature notes, positioning notes, and the comparison question.

Tool suggestion

AI assistant plus a comparison table in a document or spreadsheet.

Copy-paste prompt

You are helping me organize a competitor scan from provided notes.

Comparison goal:
[What am I trying to understand?]

Competitor or alternative notes:
[Paste notes with source labels]

Please produce:
1. Comparison table
2. Positioning summary for each competitor
3. Similarities
4. Differences
5. Possible opportunities or gaps
6. Claims that need source-checking
7. Open questions for human research

Rules:
- Use only the notes provided.
- Do not invent prices, features, customers, rankings, or results.
- Mark anything unsupported as "Not confirmed."
- Keep recommendations cautious and practical.

Expected output

A comparison table, positioning summary, differences, opportunities, and a list of claims to verify.

Human review point

Before using the scan in strategy, sales, content, or public claims, verify details against original sources.

Risk level

Medium, because competitor comparisons can easily become inaccurate or unfair if not checked.

How to improve it after the first run

Add source labels, narrow the comparison criteria, and remove any output that sounds like a claim you cannot verify.

Beginner workflow 5: Weekly planning assistant

When to use it

Use this at the start of the week to organize priorities, loose tasks, meetings, and follow-ups into a plan you can review.

Inputs needed

Task list, calendar notes, deadlines, current projects, unresolved follow-ups, and constraints like available time.

Tool suggestion

AI assistant plus your existing planner, document, project tool, or calendar. Do not let AI move meetings or message people automatically.

Copy-paste prompt

You are helping me organize a weekly plan for human review.

Week context:
[Describe the week, constraints, or focus]

Tasks and notes:
[Paste tasks, deadlines, meetings, follow-ups, and project notes]

Please produce:
1. Top priorities
2. Suggested weekly plan by day
3. Tasks that need clarification
4. Risks or overloaded days
5. Follow-ups I should not forget
6. Items that should be delegated, delayed, or removed

Rules:
- Do not invent deadlines or commitments.
- Do not schedule meetings or send messages.
- Mark uncertain items as "Needs confirmation."
- Keep the plan realistic and easy to edit.

Expected output

A practical draft plan with priorities, day-by-day suggestions, unclear items, risks, and follow-ups.

Human review point

Before using the plan, confirm deadlines, calendar conflicts, energy limits, and whether anything important is missing.

Risk level

Low for personal planning. Medium if the plan affects a team or client commitments.

How to improve it after the first run

Add better constraints, ask for shorter plans, include time estimates, and note what the AI consistently over- or under-prioritizes.

Mini exercise

Choose one of the five workflows and run it with a realistic but safe input. Highlight every uncertain item. Then write one change you would make to the prompt before using it again.

Workflow 1: Market research

Who it is for

Founders, marketers, creators, consultants, and AI companies exploring an audience, category, product idea, or market angle.

When to use it

Use this before creating a campaign, choosing a product angle, writing a landing page, or deciding which audience problem to investigate further.

Inputs needed

Customer notes, survey responses, public reports, forum excerpts, sales call notes, competitor notes, and the specific research question.

Suggested tools

AI assistant, research document, spreadsheet for themes, and original source links for verification.

Step-by-step workflow

• Collect 3 to 8 source excerpts or notes.
• Label each source clearly.
• Ask AI to summarize patterns and separate facts from assumptions.
• Review the original sources for important claims.
• Save the final brief and open questions.

Copy-paste prompt

You are helping me organize market research from provided source material only.

Research question:
[What market, audience, or problem am I trying to understand?]

Source material:
[Paste labeled excerpts, notes, survey responses, or research snippets]

Please produce:
1. Short answer to the research question
2. Key audience segments or user types mentioned
3. Repeated problems, desires, objections, or buying triggers
4. Evidence from the source material for each theme
5. Assumptions or interpretations
6. Claims that need verification
7. Open questions for follow-up research

Rules:
- Use only the source material provided.
- Do not invent market size, statistics, sources, or trends.
- Label unsupported items as "Not confirmed."
- Keep the output practical for a founder or marketer.

Expected output

A concise research brief with themes, evidence, assumptions, verification needs, and next research questions.

Human review point

Before using the brief for strategy, content, product decisions, or sales messaging, verify the source material and remove unsupported claims.

Risk level

Medium. Market research can influence business decisions, so weak sources or invented claims can lead you in the wrong direction.

Common failure mode

AI turns a small sample of notes into broad market conclusions.

How to verify the output

Check each major theme against the original source excerpts. Look for overgeneralized language such as “everyone,” “the market,” or “most customers” when the source only supports a small observation.

How to turn it into an SOP

Document the required number of sources, source-label format, prompt, output sections, verification checklist, owner, and where the approved brief is stored.

Workflow 2: Tool comparison

Who it is for

Operators, founders, marketers, creators, and AI teams comparing tools before choosing what to test.

When to use it

Use this when comparing AI tools, automation platforms, content tools, CRM add-ons, analytics tools, or workflow software.

Inputs needed

Official feature notes, pricing page notes, documentation excerpts, user requirements, constraints, and must-have features.

Suggested tools

AI assistant, spreadsheet, official vendor pages, product docs, and a human-run trial or demo when possible.

Step-by-step workflow

• Define the job the tool needs to do.
• Collect source notes from official pages or docs.
• Ask AI to create a comparison table and highlight unknowns.
• Verify pricing, features, limits, and terms directly.
• Choose what to test, not what to buy blindly.

Copy-paste prompt

You are helping me compare tools from provided notes only.

Use case:
[What job do I need the tool to do?]

Requirements:
[Must-haves, nice-to-haves, budget, team size, privacy needs, integrations]

Tool notes:
[Paste labeled notes from official pages, documentation, or demos]

Please produce:
1. Comparison table
2. Best fit for each requirement
3. Missing information I need to verify
4. Possible risks or tradeoffs
5. Questions to ask before choosing
6. Suggested next test for each tool

Rules:
- Do not invent features, prices, limits, or integrations.
- Mark anything not supported by my notes as "Not confirmed."
- Prefer cautious language.
- Do not recommend a purchase without a human test.

Expected output

A comparison table with requirements, tool fit, gaps, tradeoffs, and next tests.

Human review point

Before buying, recommending, or implementing a tool, check official pages, current pricing, privacy terms, and whether the tool works in your real workflow.

Risk level

Medium. Tool decisions can affect spend, data access, and team process.

Common failure mode

AI invents or assumes features because one tool sounds similar to another.

How to verify the output

Open the official product pages or docs and confirm every feature, price, integration, and limitation before using it in a decision.

How to turn it into an SOP

Create a repeatable comparison template with required sources, decision criteria, verification steps, trial notes, and final human approver.

Workflow 3: Source checking

Who it is for

Creators, marketers, editors, researchers, founders, and AI companies preparing public content, reports, or claims.

When to use it

Use this before publishing content, making product claims, comparing tools, citing statistics, or using research in a sales or strategy document.

Inputs needed

Draft content, claim list, source links or excerpts, publication date if known, and the level of proof needed.

Suggested tools

AI assistant for organizing claims, browser for source verification, source document, and a checklist.

Step-by-step workflow

• Paste the draft or claim list.
• Ask AI to extract claims that need verification.
• Match each claim to a source or label it unsupported.
• Check source quality and date manually.
• Revise or remove unsupported claims.

Copy-paste prompt

You are helping me source-check a draft.

Draft or claim list:
[Paste draft or claims]

Available sources:
[Paste source links, excerpts, or notes]

Please produce:
1. Claims that need verification
2. Source matched to each claim, if provided
3. Claims not supported by the provided sources
4. Claims that may be outdated
5. Claims that sound exaggerated or too broad
6. Safer wording options

Rules:
- Do not invent sources.
- Do not say a claim is verified unless the provided source supports it.
- Label unsupported claims as "Needs source."
- Label date-sensitive claims as "Check current source."

Expected output

A claim review table showing supported, unsupported, outdated, and exaggerated claims.

Human review point

Before publishing or sending, manually open the sources and confirm that the wording is accurate and current.

Risk level

Medium for ordinary public content. High if claims involve legal, medical, financial, regulatory, or safety topics.

Common failure mode

AI says a source supports a claim when the source only loosely relates to it.

How to verify the output

Read the relevant source section yourself. Check date, author, source quality, context, and whether the claim is narrower than your wording.

How to turn it into an SOP

Create a claim-checking checklist: claim, source, source date, support level, revised wording, reviewer, and approval status.

Workflow 4: Trend scanning

Who it is for

Creators, marketers, founders, community builders, and AI companies watching changes in audience behavior, tools, topics, or industry language.

When to use it

Use this when you want a structured scan of notes from newsletters, reports, social posts, customer conversations, product launches, or community discussions.

Inputs needed

Dated notes, source excerpts, links for human reference, topic boundaries, and what counts as relevant.

Suggested tools

AI assistant, spreadsheet or database, source folder, and a recurring review cadence.

Step-by-step workflow

• Collect dated source notes over a defined period.
• Ask AI to group patterns and separate signal from speculation.
• Flag time-sensitive claims.
• Verify examples and dates manually.
• Save trends as watchlist items, not facts carved in stone.

Copy-paste prompt

You are helping me scan for trends from provided notes only.

Topic boundary:
[What area am I scanning?]

Time period:
[Dates covered by these notes]

Source notes:
[Paste dated notes or excerpts]

Please produce:
1. Possible trends or repeated signals
2. Evidence from the notes for each trend
3. What is confirmed vs. speculative
4. Time-sensitive claims to verify
5. Counter-signals or exceptions
6. Follow-up questions
7. Suggested watchlist for next scan

Rules:
- Do not invent sources or current facts.
- Do not call something a trend from one weak example.
- Label speculation clearly.
- Keep the output useful for planning, not hype.

Expected output

A trend scan with possible patterns, evidence, speculation labels, counter-signals, and watchlist items.

Human review point

Before using a trend in strategy or content, check whether the evidence is broad enough and current enough.

Risk level

Medium. Trend scans can lead to strategic overreaction if the source base is weak.

Common failure mode

AI turns a few loud examples into a confident trend narrative.

How to verify the output

Check source dates, number of independent examples, source credibility, and whether there are counterexamples.

How to turn it into an SOP

Define scan frequency, source list, required date labels, trend evidence threshold, reviewer, and where watchlist items are stored.

Workflow 5: YouTube video research

Who it is for

Creators, producers, marketers, founders, and AI companies making educational or product-led video content.

When to use it

Use this before writing a video outline, planning a tutorial, or deciding what angle to take on a topic.

Inputs needed

Topic, audience, viewer questions, source notes, transcript excerpts, competitor video observations, and claims to verify.

Suggested tools

AI assistant, research document, transcript notes, YouTube notes collected manually, and source links for verification.

Step-by-step workflow

• Define the viewer and the question the video should answer.
• Collect source notes and competitor observations.
• Ask AI to organize angles, audience questions, and verification needs.
• Review for originality, accuracy, and usefulness.
• Use the brief to write a human-reviewed outline.

Copy-paste prompt

You are helping me prepare research for a YouTube video.

Video topic:
[Topic]

Target viewer:
[Who is this for?]

Viewer question:
[What should the video help them understand or do?]

Source notes and transcript excerpts:
[Paste notes]

Competitor or similar video observations:
[Paste observations, not copied scripts]

Please produce:
1. Research brief
2. Viewer questions and pain points
3. Possible video angles
4. Useful examples or demonstrations
5. Claims that need verification
6. Gaps in the research
7. Suggested outline direction

Rules:
- Do not invent facts, sources, or statistics.
- Do not copy competitor structure too closely.
- Mark unsupported claims as "Needs source."
- Keep the brief practical for creating an original video.

Expected output

A practical video research brief with viewer questions, possible angles, examples, claims to verify, and outline direction.

Human review point

Before scripting, verify claims, check originality, and decide whether the angle genuinely serves the viewer.

Risk level

Medium if the video is public or makes product, performance, legal, financial, or technical claims.

Common failure mode

AI produces a generic video angle or repeats competitor framing without adding a clear point of view.

How to verify the output

Check source links, compare against your own expertise, verify any claims, and make sure the outline is not just a copy of existing videos.

How to turn it into an SOP

Create a repeatable video research brief template with required sources, viewer question, competitor scan notes, verification checklist, and outline approval step.

Workflow 6: Competitor positioning scan

Who it is for

Founders, marketers, product teams, consultants, and AI companies comparing positioning, messaging, and category language.

When to use it

Use this before writing positioning, updating a landing page, preparing a sales narrative, or reviewing sponsor/category fit.

Inputs needed

Competitor homepage excerpts, pricing notes if verified, product descriptions, feature notes, audience claims, and your own positioning goal.

Suggested tools

AI assistant, spreadsheet, positioning document, official competitor pages, and manual source checks.

Step-by-step workflow

• Collect excerpts from official competitor pages.
• Ask AI to summarize positioning and claims from provided notes.
• Compare audience, promise, proof, pricing notes, and differentiators.
• Flag claims that need source checking.
• Use the scan to inform human positioning work.

Copy-paste prompt

You are helping me scan competitor positioning from provided notes only.

Our positioning question:
[What are we trying to understand?]

Competitor notes:
[Paste labeled homepage, product, pricing, or messaging notes]

Please produce:
1. Positioning summary for each competitor
2. Audience each competitor appears to target
3. Main promise or value proposition
4. Proof points or claims used
5. Differentiators and similarities
6. Messaging gaps or opportunities
7. Claims that need verification
8. Questions for follow-up research

Rules:
- Use only the notes provided.
- Do not invent pricing, customers, results, funding, rankings, or market position.
- Mark unsupported claims as "Not confirmed."
- Keep recommendations cautious and useful.

Expected output

A positioning scan with competitor summaries, audience signals, claims, similarities, gaps, and follow-up questions.

Human review point

Before using the scan in strategy or public messaging, verify competitor claims and avoid unfair or unsupported comparisons.

Risk level

Medium. Competitive claims can be misleading if source quality is weak or outdated.

Common failure mode

AI invents differentiators or turns marketing language into factual claims.

How to verify the output

Check official competitor pages, current docs, pricing pages, and any claims you plan to repeat or respond to.

How to turn it into an SOP

Document source collection rules, comparison fields, verification requirements, reviewer, and update frequency.

Workflow 7: Customer pain-point research

Who it is for

Creators, marketers, founders, customer success teams, product teams, and AI companies learning what users struggle with.

When to use it

Use this when organizing interviews, support tickets, survey responses, reviews, community posts, or sales notes into themes.

Inputs needed

Customer quotes or notes, support snippets, interview summaries, survey answers, segment labels if available, and privacy-safe context.

Suggested tools

AI assistant, spreadsheet or tagging document, approved customer notes, and privacy review before pasting data.

Step-by-step workflow

• Remove unnecessary personal information.
• Label sources by segment or context when useful.
• Ask AI to group pain points and quote evidence.
• Separate direct customer language from interpretation.
• Review themes before using them in messaging or product decisions.

Copy-paste prompt

You are helping me organize customer pain-point research from provided notes.

Customer segment or context:
[Who are these notes from?]

Customer notes, quotes, or support snippets:
[Paste privacy-safe notes]

Please produce:
1. Repeated pain-point themes
2. Supporting quotes or notes for each theme
3. Customer language worth preserving
4. Possible causes or interpretations
5. Assumptions to verify
6. Product, content, or sales questions this raises
7. Privacy concerns or details to remove before sharing

Rules:
- Do not invent customer quotes.
- Separate direct customer language from interpretation.
- Do not overgeneralize from a small sample.
- Mark uncertain themes as "Needs more evidence."

Expected output

A pain-point brief with themes, supporting quotes, customer language, assumptions, and follow-up questions.

Human review point

Before using the findings in messaging, product decisions, or public content, confirm that the themes are supported and privacy-safe.

Risk level

Medium because customer data and customer interpretation can affect trust, messaging, and product decisions.

Common failure mode

AI overgeneralizes from a small number of quotes or turns one loud complaint into a broad customer truth.

How to verify the output

Check the original notes, count how many independent examples support each theme, and remove private or identifying details before sharing.

How to turn it into an SOP

Create a repeatable process for anonymizing notes, tagging sources, grouping themes, preserving quotes, reviewing privacy, and approving shared insights.

Mini exercise

Choose one research workflow from this module. Gather three source excerpts, label them, run the prompt, and highlight every claim that needs verification. Then write one SOP rule that would make the workflow safer next time.

Workflow 1: Blog outline

Who it is for

Creators, marketers, founders, AI educators, and teams writing practical articles for beginners.

When to use it

Use this before drafting a blog post from research notes, a video transcript, a workflow example, or a course lesson.

Inputs needed

Topic, target reader, source notes, original opinion, examples, claims to verify, and the intended takeaway.

Suggested tools

AI assistant, writing document, source notes, and a checklist for claims and originality.

Step-by-step workflow

• Define the reader and the practical problem.
• Paste source notes and first-hand examples.
• Ask AI for an outline, not a finished article.
• Add your opinion, examples, tests, or screenshots.
• Review the outline for usefulness before drafting.

Copy-paste prompt

You are helping me create a practical blog outline for Kingy AI's audience.

Topic:
[Topic]

Reader:
[Creator, marketer, founder, operator, AI-curious professional, etc.]

Source notes:
[Paste notes, transcript excerpts, research, or workflow details]

First-hand perspective to include:
[My opinion, example, observation, test, comparison, or case detail]

Please produce:
1. Working title options
2. Practical promise of the article
3. Section-by-section outline
4. Real examples to include
5. Claims that need verification
6. Places where first-hand perspective is needed
7. Suggested conclusion and next action

Rules:
- Do not write generic AI hype.
- Do not invent examples, statistics, or results.
- Keep the outline useful for a smart non-technical reader.
- Mark weak or unsupported sections as "Needs stronger source."

Expected output

A structured outline with a clear reader, practical sections, source-based claims, and places for original perspective.

Human review point

Before drafting, check whether the outline teaches something specific and whether every claim has a source or example.

Risk level

Medium if the article will be public. Claims, examples, and recommendations need review.

Common failure mode

The outline sounds polished but could apply to any AI blog.

Quality checklist

• Does the outline solve a specific problem?
• Does it include real examples or observations?
• Are claims marked for verification?
• Would the reader know what to do next?

How to make the output sound less generic

Add a specific audience, a real workflow example, your opinion on what beginners should avoid, and a concrete before/after.

How to add first-hand perspective

Add notes from a tool test, a screenshot description, a real prompt you used, or a short case study from a creator, marketer, founder, or operator.

Workflow 2: YouTube description

Who it is for

YouTube creators, AI educators, course creators, and marketers publishing practical videos.

When to use it

Use this after the video is outlined, recorded, or transcribed, not before the actual content exists.

Inputs needed

Video title, transcript or outline, key links, audience, main takeaway, disclaimers, and calls to action.

Suggested tools

AI assistant, transcript, YouTube upload draft, and link checklist.

Step-by-step workflow

• Paste the transcript or approved outline.
• Ask AI to summarize the real video, not invent extras.
• Generate description, chapters, and pinned comment options.
• Check links, claims, and timestamps manually.
• Edit for voice and clarity before publishing.

Copy-paste prompt

You are helping me draft a YouTube description from source material.

Video title:
[Title]

Audience:
[Who is this for?]

Transcript or outline:
[Paste transcript or outline]

Links or resources to include:
[Paste approved links or say none]

Please produce:
1. Short YouTube description
2. Longer description with practical summary
3. Chapter suggestions if the source supports them
4. Pinned comment option
5. Claims, links, or timestamps I must verify

Rules:
- Use only the transcript or outline.
- Do not invent links, timestamps, tools, or claims.
- Avoid exaggerated promises.
- Keep the description useful and clear.

Expected output

A description draft, optional chapters, pinned comment, and verification list.

Human review point

Before publishing, verify links, timestamps, claims, and whether the description accurately represents the video.

Risk level

Medium because the content is public.

Common failure mode

AI adds claims, tools, or promises that are not actually in the video.

Quality checklist

• Does the description match the video?
• Are links correct?
• Are timestamps real?
• Is the call to action honest?

How to make the output sound less generic

Add the exact workflow, tool test, or lesson shown in the video.

How to add first-hand perspective

Mention what you demonstrated, what surprised you, or what a beginner should watch for.

Workflow 3: Thumbnail concepts

Who it is for

YouTube creators, designers, marketers, and educators brainstorming visual directions.

When to use it

Use this after the video angle is clear and you know what promise the thumbnail must communicate.

Inputs needed

Video topic, audience, main tension, visual assets available, brand constraints, and thumbnail examples you like or dislike.

Suggested tools

AI assistant for concepting, design tool for execution, and human review for clarity and accuracy.

Step-by-step workflow

• Define the video promise and viewer tension.
• Ask AI for thumbnail concepts, not finished design.
• Review for clarity, accuracy, and non-clickbait wording.
• Choose 2 to 3 concepts for design exploration.
• Test whether the thumbnail matches the video.

Copy-paste prompt

You are helping me brainstorm YouTube thumbnail concepts.

Video topic:
[Topic]

Viewer:
[Who is this for?]

Main tension or promise:
[What problem or contrast should the thumbnail communicate?]

Brand or visual constraints:
[Colors, style, face/no face, screenshots, tools, etc.]

Please produce:
1. 8 thumbnail concept ideas
2. Suggested short text for each, 1 to 4 words
3. Visual composition notes
4. What the concept communicates
5. Risks of misleading or overpromising

Rules:
- Do not create clickbait that the video does not support.
- Avoid exaggerated income, guarantee, or miracle claims.
- Keep concepts clear at small size.
- Include ideas that could use real screenshots or workflow artifacts.

Expected output

A set of thumbnail concepts with text, composition, meaning, and risk notes.

Human review point

Before designing, check whether the thumbnail promise is true and whether it matches the video content.

Risk level

Medium because misleading thumbnails damage trust.

Common failure mode

AI suggests dramatic concepts that overpromise what the video delivers.

Quality checklist

• Is the concept readable at small size?
• Does it match the actual video?
• Does it avoid fake urgency or false claims?
• Can it include a real screenshot or artifact?

How to make the output sound less generic

Use specific visual artifacts: a workflow canvas, prompt window, checklist, before/after output, or tool comparison table.

How to add first-hand perspective

Use a real screenshot, test result, or visual from the workflow demonstrated in the video.

Workflow 4: Newsletter draft

Who it is for

Creators, founders, marketers, consultants, and AI educators sending practical updates to an audience.

When to use it

Use this when you have a source idea, lesson, tool test, workflow example, or weekly observation to turn into a draft.

Inputs needed

Main idea, source notes, audience, tone, first-hand observation, links, and the desired call to action.

Suggested tools

AI assistant, writing document, newsletter platform draft, and link/source checklist.

Step-by-step workflow

• Write the real point you want to make.
• Paste source notes and first-hand observations.
• Ask AI for a draft and subject line options.
• Edit heavily for voice, specificity, and usefulness.
• Verify claims and links before sending.

Copy-paste prompt

You are helping me draft a practical newsletter for Kingy AI's audience.

Audience:
[Who receives this?]

Main point:
[What should readers understand or do?]

Source notes:
[Paste notes, lesson, workflow, observation, or test result]

First-hand perspective:
[What I personally observed, tested, recommend, or disagree with]

Call to action:
[What should readers do next?]

Please produce:
1. 5 subject line options
2. Short newsletter draft
3. Practical takeaway section
4. Optional checklist or prompt snippet
5. Claims or links to verify

Rules:
- Do not invent personal experience.
- Avoid generic AI hype.
- Keep the tone practical and direct.
- Mark unsupported claims as "Needs source."

Expected output

Subject lines, a draft newsletter, practical takeaway, optional checklist, and verification notes.

Human review point

Before sending, check voice, claims, links, examples, and whether the email respects the reader’s time.

Risk level

Medium because it is sent to an audience and may affect trust.

Common failure mode

The draft sounds polished but says nothing specific from your real work.

Quality checklist

• Does it have a clear point?
• Is there a real example or observation?
• Are links and claims checked?
• Would the reader find it useful without buying anything?

How to make the output sound less generic

Add one opinion, one example, and one practical artifact such as a checklist or prompt.

How to add first-hand perspective

Describe what you tested, what failed, what you changed, or what you would do differently next time.

Workflow 5: LinkedIn post

Who it is for

Founders, consultants, creators, marketers, and AI practitioners sharing useful work observations.

When to use it

Use this when you have one clear lesson from a project, workflow, tool test, customer question, or content piece.

Inputs needed

Core idea, audience, personal observation, specific example, proof or caveat, and desired discussion prompt.

Suggested tools

AI assistant for variants, writing document, and human editing for voice.

Step-by-step workflow

• Write the point in one sentence.
• Add the real example or observation.
• Ask AI for post structures and hooks.
• Remove hype and generic phrasing.
• Verify claims and publish only after editing.

Copy-paste prompt

You are helping me draft LinkedIn post options from a real work observation.

Audience:
[Who is this for?]

Core idea:
[One sentence]

Real example or observation:
[What happened, what I tested, what I noticed, or what I learned]

Tone:
[Direct, practical, thoughtful, etc.]

Please produce:
1. 5 hook options
2. 3 post drafts with different structures
3. A version that is shorter and more direct
4. Generic phrases to remove
5. Claims or examples that need verification

Rules:
- Do not invent personal stories.
- Do not exaggerate results.
- Avoid "game changer" style hype.
- Keep the post useful for people doing real work with AI.

Expected output

Hook options, draft variants, a shorter version, generic phrase warnings, and verification notes.

Human review point

Before posting, check whether the example is true, the opinion is yours, and the wording does not overstate the lesson.

Risk level

Medium because it is public and attached to personal or brand trust.

Common failure mode

AI writes a post that sounds like every other AI productivity post.

Quality checklist

• Does it include a real observation?
• Is the hook honest?
• Does it avoid exaggerated claims?
• Would a smart reader learn something concrete?

How to make the output sound less generic

Use plain language, a specific scenario, and a clear opinion about what worked or failed.

How to add first-hand perspective

Include what you saw while testing a prompt, building a workflow, comparing tools, or reviewing AI output.

Workflow 6: Short-form repurposing

Who it is for

Creators, marketers, editors, and founders turning long-form content into short clips, posts, or snippets.

When to use it

Use this after you have a transcript, article, podcast, webinar, or lesson that already contains real substance.

Inputs needed

Transcript or article, audience, platform, tone, key message, and any claims to avoid.

Suggested tools

AI assistant, transcript editor, content calendar, and human review for accuracy and platform fit.

Step-by-step workflow

• Paste the source material.
• Ask AI to extract moments, not invent new points.
• Generate short-form options for the target platform.
• Edit for voice and accuracy.
• Verify quotes, claims, and context before publishing.

Copy-paste prompt

You are helping me repurpose long-form source material into short-form content.

Source material:
[Paste transcript, article, lesson, or notes]

Platform:
[LinkedIn, YouTube Shorts, TikTok, Instagram, X, etc.]

Audience:
[Who is this for?]

Please produce:
1. 10 short-form content ideas grounded in the source
2. Hook for each idea
3. Source moment or quote it came from
4. Draft caption or script
5. Claims or context that need verification

Rules:
- Use only the source material.
- Do not invent quotes or results.
- Preserve the meaning of the original.
- Avoid generic AI productivity advice unless the source specifically supports it.

Expected output

Short-form ideas, hooks, source references, draft captions/scripts, and verification notes.

Human review point

Before publishing, confirm that the short-form version does not distort the original point.

Risk level

Medium because short content can easily remove context.

Common failure mode

AI creates catchy snippets that misrepresent the source material.

Quality checklist

• Does each idea map to a real source moment?
• Is the hook accurate?
• Is context preserved?
• Are claims verified?

How to make the output sound less generic

Anchor each piece to a real line, example, workflow, or mistake from the source.

How to add first-hand perspective

Add a short note explaining why the moment matters or what you learned from it.

Workflow 7: Content refresh workflow

Who it is for

Site owners, marketers, creators, SEO writers, AI educators, and teams maintaining older content.

When to use it

Use this when an old article, guide, or landing page may be outdated, thin, generic, or missing practical examples.

Inputs needed

Existing content, update goal, current source notes, first-hand updates, outdated sections, and target reader.

Suggested tools

AI assistant, CMS draft, source-checking workflow, and human editor.

Step-by-step workflow

• Paste the existing content.
• Ask AI to identify outdated, generic, or unsupported sections.
• Add updated source material and first-hand observations.
• Ask for a refresh plan, not an automatic rewrite.
• Human edits and verifies before publishing.

Copy-paste prompt

You are helping me refresh existing content without making unsupported changes.

Existing content:
[Paste content]

Refresh goal:
[Improve accuracy, usefulness, examples, SEO, clarity, conversion, etc.]

Updated source notes or first-hand observations:
[Paste notes]

Please produce:
1. Sections that are still useful
2. Sections that seem outdated or weak
3. Claims that need source-checking
4. Places to add real examples, screenshots, tests, comparisons, or case studies
5. Suggested refresh outline
6. Edits to avoid because they would require unsupported claims

Rules:
- Do not invent new facts.
- Do not rewrite the full piece yet.
- Mark uncertain claims as "Needs verification."
- Prioritize usefulness over adding more words.

Expected output

A refresh plan identifying what to keep, update, verify, strengthen, and avoid.

Human review point

Before updating the live page, verify facts, links, examples, and whether changes match the brand voice.

Risk level

Medium if public content is updated. Higher if the content covers sensitive topics.

Common failure mode

AI adds new sections that sound helpful but are not supported by current sources or first-hand experience.

Quality checklist

• Were outdated claims identified?
• Were unsupported additions avoided?
• Were real examples added?
• Does the refreshed piece help the reader more than before?

How to make the output sound less generic

Add updated screenshots, recent tests, workflow examples, and specific lessons learned.

How to add first-hand perspective

Explain what changed since the original piece and what you now recommend based on actual use.

Workflow 8: Content quality review

Who it is for

Anyone publishing AI-assisted content: creators, marketers, founders, editors, and AI companies.

When to use it

Use this before publishing or sending any AI-assisted draft.

Inputs needed

Draft content, target reader, source material, claims, brand voice notes, and desired action.

Suggested tools

AI assistant for review, source-checking workflow, human editor, and publishing checklist.

Step-by-step workflow

• Paste the draft and intended audience.
• Ask AI to review for generic phrasing, unsupported claims, and missing first-hand perspective.
• Ask for a source and quality checklist.
• Human verifies every important claim.
• Revise before publishing.

Copy-paste prompt

You are reviewing this AI-assisted content for quality before publication.

Draft:
[Paste draft]

Audience:
[Who is this for?]

Source material:
[Paste notes or say what sources were used]

Please review for:
1. Generic or vague sections
2. Unsupported claims
3. Missing first-hand perspective
4. Places to add examples, screenshots, tests, comparisons, or case studies
5. Tone issues
6. Privacy or trust concerns
7. Specific edits to make the draft more useful

Rules:
- Do not rewrite the full draft.
- Do not invent facts.
- Mark claims as "Needs source" when support is missing.
- Prioritize usefulness, accuracy, and originality.

Expected output

A quality review with generic sections, unsupported claims, missing perspective, trust risks, and concrete edits.

Human review point

Before publishing, decide which edits to make, verify claims, and confirm the final piece reflects your actual point of view.

Risk level

Medium for public content. High if the draft includes sensitive advice or regulated topics.

Common failure mode

AI gives compliments instead of a useful critique, or rewrites the draft into more generic language.

Quality checklist

• Is the piece useful for a specific reader?
• Are claims verified?
• Does it include first-hand evidence?
• Does it avoid generic AI language?
• Does it have a clear next step?

How to make the output sound less generic

Ask for critique, not polish. Then add real examples, specific language, and actual experience.

How to add first-hand perspective

Add what you tested, what surprised you, what failed, what you recommend, and what you would avoid.

Mini exercise

Choose one piece of source material: a transcript, lesson, workflow, test note, or old article. Run one workflow from this module. Then add at least one first-hand observation and remove at least three generic phrases before considering publication.

Workflow 1: ICP research

Who it is for

AI startups, creators, founders, consultants, and marketers defining their ideal customer profile.

When to use it

Use this before writing messaging, choosing a channel, building a lead list, or planning sponsor outreach.

Inputs needed

Customer notes, sales call notes, support themes, audience data, product use cases, testimonials if real, and constraints around who you do not serve.

Suggested tools

AI assistant, CRM or spreadsheet, customer notes document, and source-checking workflow.

Step-by-step workflow

• Gather real customer or audience notes.
• Ask AI to group patterns and separate evidence from assumptions.
• Draft ICP segments with pains, triggers, use cases, and disqualifiers.
• Review against real customers and revenue, not just what sounds good.
• Save a first ICP draft and open questions.

Copy-paste prompt

You are helping me draft an ideal customer profile from provided evidence.

Product, service, or media property:
[Describe it]

Source notes:
[Paste customer notes, audience notes, sales notes, support themes, or real examples]

Please produce:
1. Possible ICP segments
2. Pain points for each segment
3. Buying or interest triggers
4. Jobs they are trying to get done
5. Reasons they may not be a fit
6. Evidence from the notes for each segment
7. Assumptions to verify
8. Questions to ask in future research

Rules:
- Use only the source notes.
- Do not invent customer types, revenue, or urgency.
- Mark weakly supported segments as "Needs more evidence."
- Keep the output practical for messaging and research.

Expected output

ICP segment drafts with pains, triggers, fit criteria, evidence, assumptions, and follow-up questions.

Human review point

Before using the ICP in campaigns or sales targeting, confirm it with real customer evidence and business context.

Risk level

Medium. ICP assumptions can shape messaging, lead lists, and product direction.

Common failure mode

AI creates neat customer segments that are not supported by real data.

How to personalize without being creepy

Personalize around role, public business context, stated problem, or relevant use case. Do not use sensitive personal details or pretend to know private motivations.

What must be reviewed by a human before sending

Segment fit, claims about the customer’s needs, any implied pain point, and whether the message is relevant enough to justify contact.

Workflow 2: Sponsor lead research

Who it is for

Creators, newsletter operators, YouTube channels, podcast hosts, AI media companies, and partnership teams.

When to use it

Use this before deciding whether a sponsor is a good fit for Kingy AI’s audience or drafting a partnership note.

Inputs needed

Public sponsor notes, product page excerpts, audience fit criteria, past sponsor guidelines, approved media kit notes, and any public contact context.

Suggested tools

AI assistant, sponsor spreadsheet, public company pages, media kit, and human source verification.

Step-by-step workflow

• Collect public information and approved internal fit criteria.
• Ask AI to summarize fit, concerns, and open questions.
• Check for audience relevance and brand safety.
• Draft outreach only if the fit is plausible.
• Human reviews before contacting anyone.

Copy-paste prompt

You are helping me research a potential sponsor lead for a practical AI education audience.

Sponsor or company:
[Name]

Public notes and source excerpts:
[Paste notes]

Audience fit criteria:
[Describe the audience and sponsor standards]

Please produce:
1. Sponsor fit summary
2. Why this might be relevant to the audience
3. Possible concerns or misfit signals
4. Public claims that need verification
5. Outreach angle options
6. Questions to answer before contacting
7. A respectful draft note for human review only

Rules:
- Do not invent budget, contacts, results, customers, or sponsorship interest.
- Do not imply endorsement.
- Keep personalization based on public, relevant business context.
- Do not send anything.

Expected output

A sponsor research brief with fit, concerns, verification needs, possible outreach angles, and a draft note for review.

Human review point

Before outreach, verify the company, audience fit, claims, contact relevance, and whether sponsorship alignment is real.

Risk level

Medium because it affects business relationships and public trust.

Common failure mode

AI exaggerates sponsor fit or writes an outreach note that sounds like a mass template.

How to personalize without being creepy

Reference public product relevance, audience overlap, or a specific approved fit reason. Do not mention personal browsing, private assumptions, or hidden data.

What must be reviewed by a human before sending

Sponsor fit, claims about audience, contact choice, tone, offer, and whether the note is honest about why you are reaching out.

Workflow 3: Landing page teardown

Who it is for

Founders, marketers, AI startups, creators selling products, and consultants reviewing landing pages.

When to use it

Use this before rewriting a landing page, improving conversion clarity, or comparing messaging against an ICP.

Inputs needed

Landing page copy, target audience, offer details, traffic source, current concerns, and any analytics or customer feedback you are allowed to use.

Suggested tools

AI assistant, page copy document, analytics notes if available, and human conversion review.

Step-by-step workflow

• Paste the page copy and state the target reader.
• Ask AI to review clarity, promise, proof, friction, and missing context.
• Ask for questions, not just recommendations.
• Review suggestions against real audience and offer constraints.
• Prioritize changes for a human rewrite.

Copy-paste prompt

You are helping me review a landing page for clarity and relevance.

Target audience:
[Who is this page for?]

Offer:
[What is being offered?]

Page copy:
[Paste copy]

Please produce:
1. One-sentence page promise
2. What is clear
3. What is confusing
4. Missing proof or missing context
5. Claims that need evidence
6. Friction or trust issues
7. Questions a visitor may still have
8. Suggested improvements, prioritized

Rules:
- Do not invent conversion data.
- Do not make guaranteed result claims.
- Keep recommendations practical.
- Mark unsupported claims as "Needs proof."

Expected output

A teardown focused on clarity, relevance, proof, friction, questions, and prioritized improvements.

Human review point

Before changing the page, confirm the recommendations match the real offer, audience, analytics, and brand standards.

Risk level

Medium because landing page changes can affect revenue and trust.

Common failure mode

AI suggests generic conversion advice that ignores the actual audience or offer.

How to personalize without being creepy

Use audience-level relevance such as role, goal, and use case. Do not imply you know a visitor’s private behavior.

What must be reviewed by a human before sending

Any revised claims, pricing references, testimonials, guarantees, proof points, and calls to action.

Workflow 4: Offer positioning

Who it is for

Founders, creators, consultants, AI startups, course sellers, and marketers shaping an offer.

When to use it

Use this when an offer is real but the messaging feels vague, too broad, or too similar to competitors.

Inputs needed

Offer details, audience notes, customer pains, proof you actually have, competitor notes, constraints, and what you do not want to claim.

Suggested tools

AI assistant, positioning document, customer notes, competitor scan, and human strategy review.

Step-by-step workflow

• Define the offer and target audience.
• Paste real customer and competitor notes.
• Ask AI for positioning options with caveats.
• Remove unsupported claims and vague differentiation.
• Choose a positioning direction to test with humans.

Copy-paste prompt

You are helping me explore offer positioning from provided information.

Offer:
[Describe the offer]

Target audience:
[Who it is for]

Customer notes or pains:
[Paste notes]

Proof or constraints:
[What can we honestly claim? What should we avoid claiming?]

Competitor or alternative notes:
[Paste notes if available]

Please produce:
1. Positioning summary
2. 5 possible positioning angles
3. Best-fit audience for each angle
4. Proof needed for each angle
5. Claims to avoid
6. Differentiation opportunities
7. Questions to validate with customers

Rules:
- Do not invent proof, testimonials, revenue, or outcomes.
- Avoid guaranteed results language.
- Keep positioning specific and testable.
- Mark risky claims as "Needs proof."

Expected output

Positioning options with audience fit, proof requirements, claims to avoid, and validation questions.

Human review point

Before using positioning publicly, confirm proof, customer fit, legal/compliance concerns, and whether the promise is honest.

Risk level

Medium because positioning shapes public claims and buyer expectations.

Common failure mode

AI makes the offer sound bigger, easier, or more proven than it is.

How to personalize without being creepy

Adapt positioning to public audience needs and stated use cases, not hidden personal assumptions.

What must be reviewed by a human before sending

Offer promise, proof, claims, audience fit, price/value language, and any implied outcome.

Workflow 5: Outreach draft with human review

Who it is for

Founders, creators, sponsor partnership teams, consultants, and marketers drafting respectful outreach.

When to use it

Use this only after you have a relevant reason to contact someone and enough public or permission-based context.

Inputs needed

Recipient role or company, relevant public context, reason for outreach, offer, proof you can honestly mention, and desired next step.

Suggested tools

AI assistant for draft options, CRM or spreadsheet for tracking, and email client for human-sent messages only.

Step-by-step workflow

• Confirm there is a legitimate, relevant reason to reach out.
• Provide only appropriate public or permission-based context.
• Ask AI for draft options with a respectful tone.
• Human edits for truth, relevance, brevity, and consent.
• Human sends only if the message still feels respectful.

Copy-paste prompt

You are helping me draft respectful outreach for human review.

Recipient context:
[Role, company, or public business context]

Reason this may be relevant:
[Specific, honest reason]

Offer or ask:
[What I am asking for]

Proof or credibility I can honestly mention:
[Real proof only]

Desired next step:
[Reply, short call, review a sponsor fit, etc.]

Please produce:
1. A short outreach draft
2. A warmer version
3. A more direct version
4. Personalization used and why it is appropriate
5. Claims or details a human must verify
6. A version that is easy to decline

Rules:
- Do not invent personalization.
- Do not imply a relationship that does not exist.
- Do not use pressure, fake urgency, or deceptive familiarity.
- Do not send anything.
- Keep it respectful and concise.

Expected output

Draft outreach options, personalization rationale, verification notes, and an easy-to-decline version.

Human review point

Before sending, review relevance, personalization, claims, tone, consent, and whether the message should be sent at all.

Risk level

Medium. Outreach affects trust and reputation.

Common failure mode

AI writes a message that feels personalized but is based on assumptions or irrelevant details.

How to personalize without being creepy

Use a relevant business reason, public company context, or stated interest. Avoid personal details, hidden tracking, or overly familiar language.

What must be reviewed by a human before sending

Recipient relevance, factual accuracy, personalization source, claims, links, ask, unsubscribe/decline path where appropriate, and tone.

Workflow 6: Objection mining

Who it is for

Sales teams, founders, creators selling offers, AI startups, customer success teams, and marketers improving messaging.

When to use it

Use this after you have real sales call notes, support tickets, comments, survey responses, or customer interviews.

Inputs needed

Customer language, sales notes, objections, lost deal notes, comments, and context about the offer.

Suggested tools

AI assistant, spreadsheet for objection themes, CRM notes, and human review by sales or customer owner.

Step-by-step workflow

• Remove private or unnecessary personal data.
• Paste real objections or notes.
• Ask AI to group objections and preserve customer language.
• Separate direct quotes from interpretation.
• Use findings to improve content, FAQs, and sales prep.

Copy-paste prompt

You are helping me organize customer objections from provided notes.

Offer:
[Describe the offer]

Customer or prospect notes:
[Paste privacy-safe notes]

Please produce:
1. Objection themes
2. Direct customer language for each theme
3. Likely concern behind each objection
4. What information may answer the objection
5. Messaging or content ideas
6. Assumptions to verify
7. Questions for future calls

Rules:
- Do not invent quotes.
- Separate direct customer language from interpretation.
- Do not overgeneralize from a small sample.
- Keep recommendations respectful.

Expected output

Objection themes, customer language, likely concerns, content ideas, assumptions, and future questions.

Human review point

Before using findings in sales material, confirm the themes are supported and the wording respects customer context.

Risk level

Medium because customer interpretation affects messaging and sales behavior.

Common failure mode

AI turns one objection into a broad market belief.

How to personalize without being creepy

Use objection themes to improve relevance, not to pressure individuals with their specific concerns.

What must be reviewed by a human before sending

Any response language, claims used to answer objections, customer quotes, and whether the message respects the relationship.

Workflow 7: Sales call prep

Who it is for

Founders, consultants, account owners, sponsor partnership leads, and sales teams preparing for a conversation.

When to use it

Use this before a discovery call, sponsor conversation, demo, consulting call, or renewal discussion.

Inputs needed

Lead notes, intake form, public company context, previous messages, offer details, known constraints, and call goal.

Suggested tools

AI assistant, CRM notes, call agenda document, and human review before the call.

Step-by-step workflow

• Gather relevant and appropriate context.
• Ask AI for a prep brief, questions, risks, and assumptions.
• Check every assumption before the call.
• Use the brief to guide preparation, not to script the person.
• Update notes after the call.

Copy-paste prompt

You are helping me prepare for a sales or partnership call.

Lead or partner context:
[Paste notes, intake answers, or public business context]

Offer or partnership idea:
[Describe it]

Call goal:
[Discovery, sponsor fit, demo, renewal, etc.]

Please produce:
1. Call prep brief
2. What seems known
3. Assumptions to verify
4. Useful questions to ask
5. Possible objections or concerns
6. Relevant examples or proof I can mention if true
7. Things I should avoid saying without confirmation

Rules:
- Do not invent facts about the person or company.
- Do not make promises.
- Mark assumptions clearly.
- Keep the prep respectful and useful.

Expected output

A call prep brief with known context, assumptions, questions, possible objections, proof to verify, and things to avoid.

Human review point

Before the call, review assumptions, sensitive topics, proof, and what you should not claim.

Risk level

Medium because poor prep can create awkward assumptions or false promises.

Common failure mode

AI infers too much from limited public context and creates overconfident talking points.

How to personalize without being creepy

Use personalization as context for better questions, not as a performance of research. Keep it professional and relevant.

What must be reviewed by a human before sending

Any pre-call email, agenda, claims, proof points, and assumptions about the lead’s needs.

Workflow 8: Follow-up email draft

Who it is for

Founders, creators, consultants, sales teams, sponsor partnership leads, and account owners.

When to use it

Use this after a call or conversation where you need to summarize next steps accurately.

Inputs needed

Call notes, decisions, open questions, agreed next steps, owner names if explicitly mentioned, and any approved links or resources.

Suggested tools

AI assistant, notes document, email draft, CRM or project tool after human review.

Step-by-step workflow

• Paste the meeting or call notes.
• Ask AI to draft a follow-up with uncertainty labels.
• Review all commitments, owners, dates, and promises.
• Edit for tone and relationship.
• Human sends only after approval.

Copy-paste prompt

You are helping me draft a follow-up email for human review.

Conversation context:
[Who met and why?]

Call notes:
[Paste notes]

Approved links or resources:
[Paste links or say none]

Please produce:
1. Short follow-up email draft
2. Summary of what was discussed
3. Agreed next steps
4. Open questions
5. Items that need confirmation before sending
6. A shorter version if appropriate

Rules:
- Do not invent commitments, owners, deadlines, pricing, or promises.
- Mark unclear items as "Needs confirmation."
- Keep the tone respectful and specific.
- Do not send anything.

Expected output

A follow-up email draft, summary, next steps, open questions, and confirmation list.

Human review point

Before sending, confirm every commitment, deadline, owner, link, price, claim, and tone choice.

Risk level

Medium because follow-up emails can create commitments or misunderstandings.

Common failure mode

AI invents next steps or makes the email sound more agreed-upon than the conversation really was.

How to personalize without being creepy

Reference what was actually discussed. Do not add personal details or assumptions that were not part of the conversation.

What must be reviewed by a human before sending

Commitments, next steps, deadlines, pricing, links, claims, attachments, and whether the email accurately reflects the conversation.

Mini exercise

Choose one sales or marketing workflow. Run the prompt with safe sample notes. Then highlight every claim, personalization line, and next step that must be checked before a human could send or use it.

Workflow 1: SOP creation

Who it is for

Founders, operators, agencies, assistants, creators, and small teams documenting repeatable work.

When to use it

Use this when a process lives in someone’s head, a messy note, or a long chat thread.

Inputs needed

Process notes, screenshots or tool names if safe to share, owner, exceptions, common mistakes, and what done means.

Suggested tools

AI assistant, document editor, internal wiki, project tool, and a process owner for review.

Step-by-step workflow

• Collect messy process notes.
• Ask AI to organize them into an SOP draft.
• Mark unclear steps as “Needs clarification.”
• Have the process owner review it.
• Save the approved version where the team can find it.

Copy-paste prompt

You are helping me turn messy process notes into a draft SOP.

Process name:
[Name]

Process notes:
[Paste notes]

Please produce:
1. SOP title
2. Purpose
3. When to use this SOP
4. When not to use it
5. Required inputs
6. Step-by-step process
7. Tools or documents mentioned
8. Owner or role responsible, if mentioned
9. Common mistakes or risks
10. Open questions and missing details
11. Human review checklist

Rules:
- Do not invent steps, owners, tools, or rules.
- Mark unclear items as "Needs clarification."
- Keep the SOP simple enough for a small team to use.

Expected output

A draft SOP with steps, tools, owner, risks, open questions, and a review checklist.

Human review point

The process owner must confirm step order, exceptions, tools, permissions, and what done means.

Risk level

Low for internal drafts. Medium if the SOP affects customers, billing, access, or production work.

Common failure mode

AI fills in missing process steps that were not in the notes.

How to improve the workflow

Add screenshots, examples of completed work, common exceptions, and a stronger “do not invent steps” rule.

How to document it as an SOP

Save the final SOP with owner, last updated date, review checklist, and a place to record future changes.

Workflow 2: Client onboarding checklist

Who it is for

Agencies, consultants, freelancers, creators with sponsors, and small service businesses.

When to use it

Use this when starting a new client, sponsor, consulting project, or service engagement.

Inputs needed

Offer details, client intake notes, deliverables, timeline, required access, assets needed, and communication preferences.

Suggested tools

AI assistant, checklist document, project board, shared folder, and human project owner.

Step-by-step workflow

• Paste intake and project notes.
• Ask AI to turn them into onboarding tasks.
• Separate client tasks from internal tasks.
• Review access, deadlines, and responsibilities.
• Save as a reusable checklist template.

Copy-paste prompt

You are helping me create a client onboarding checklist.

Client or project context:
[Describe the engagement]

Intake notes:
[Paste notes]

Deliverables and timeline:
[Paste details]

Please produce:
1. Internal onboarding checklist
2. Client-facing checklist
3. Assets or access needed
4. Questions to confirm
5. Risks or blockers
6. Suggested first-week plan

Rules:
- Do not invent deadlines, access requirements, or deliverables.
- Mark unclear items as "Needs confirmation."
- Keep the checklist simple and useful.

Expected output

Internal tasks, client tasks, asset requests, access needs, questions, blockers, and a first-week plan.

Human review point

Before sending anything to the client, confirm deliverables, dates, responsibilities, and access requests.

Risk level

Medium because onboarding affects client trust and project setup.

Common failure mode

AI adds tasks that sound reasonable but are not part of the actual agreement.

How to improve the workflow

Add standard deliverables, common access needs, and a rule to separate contract facts from assumptions.

How to document it as an SOP

Create a reusable onboarding SOP with trigger, checklist owner, client email template, folder setup, and review step.

Workflow 3: Support reply draft

Who it is for

Solo operators, small support teams, founders, creators with products, and agencies handling client questions.

When to use it

Use this to draft a response, not to send automatically, when a customer or client asks a question.

Inputs needed

Customer question, relevant policy, product notes, account context if appropriate, and tone guidelines.

Suggested tools

AI assistant, help docs, support inbox draft, and human support owner.

Step-by-step workflow

• Remove unnecessary private data.
• Paste the question and approved policy/source notes.
• Ask AI for a draft reply and missing information.
• Human checks policy, tone, and accuracy.
• Human sends only after review.

Copy-paste prompt

You are helping me draft a support reply for human review.

Customer question:
[Paste question]

Relevant policy or source notes:
[Paste approved policy, product notes, or help doc excerpt]

Tone:
[Friendly, direct, calm, etc.]

Please produce:
1. Short summary of the issue
2. Draft reply
3. Missing information to ask for, if needed
4. Policy or source details used
5. Items a human must verify before sending

Rules:
- Use only the policy or source notes provided.
- Do not promise refunds, timelines, fixes, or exceptions unless clearly supported.
- Mark uncertain items as "Needs confirmation."
- Do not send anything.

Expected output

A support reply draft, issue summary, missing information, source notes used, and verification list.

Human review point

Before sending, check policy, customer context, tone, commitments, and whether escalation is needed.

Risk level

Medium. Support replies affect customers and can create commitments.

Common failure mode

AI makes a promise that the business cannot support.

How to improve the workflow

Add approved response examples, escalation rules, and a stricter “do not promise” instruction.

How to document it as an SOP

Document which policies to use, what requires escalation, who approves replies, and where sent replies are logged.

Workflow 4: Task triage

Who it is for

Founders, assistants, operators, creators, agency leads, and small teams with messy task lists.

When to use it

Use this when tasks are scattered across notes, emails, chats, or project tools and need sorting.

Inputs needed

Task list, deadlines, owners if known, project context, urgency notes, and constraints.

Suggested tools

AI assistant, task manager, spreadsheet, project board, or simple document.

Step-by-step workflow

• Paste the messy task list.
• Ask AI to group, deduplicate, and flag unclear tasks.
• Human confirms priorities, deadlines, and owners.
• Move approved tasks into the task system.
• Save unclear items for follow-up.

Copy-paste prompt

You are helping me triage a messy task list.

Context:
[Project, team, or week]

Tasks and notes:
[Paste messy list]

Please produce:
1. Clean task list
2. Duplicates or related tasks
3. Suggested priority groupings
4. Owners if clearly mentioned
5. Deadlines if clearly mentioned
6. Unclear tasks that need clarification
7. Suggested next review questions

Rules:
- Do not invent owners or deadlines.
- Mark unclear tasks as "Needs clarification."
- Keep the output simple enough to move into a task manager.

Expected output

A cleaned task list with groups, duplicates, owners/deadlines if known, and clarification questions.

Human review point

Before assigning or importing tasks, confirm priority, owner, deadline, and whether the task still matters.

Risk level

Low for personal tasks. Medium if tasks affect a team, client, or deadline.

Common failure mode

AI assigns owners or urgency based on guesses.

How to improve the workflow

Add priority rules, project names, owner conventions, and examples of well-written tasks.

How to document it as an SOP

Document when triage happens, who reviews, priority labels, task destination, and how unclear tasks are handled.

Workflow 5: Project status update

Who it is for

Project leads, founders, operators, agencies, creators, and small client teams.

When to use it

Use this weekly or before a stakeholder update when you have scattered progress notes.

Inputs needed

Progress notes, completed work, blockers, risks, metrics if verified, decisions needed, and next steps.

Suggested tools

AI assistant, project board, status document, and human project owner.

Step-by-step workflow

• Gather progress notes and verified facts.
• Ask AI to draft a structured update.
• Review for accuracy, tone, and missing context.
• Confirm blockers and decisions needed.
• Send or post only after approval.

Copy-paste prompt

You are helping me draft a project status update for human review.

Project:
[Name]

Audience:
[Internal team, client, sponsor, etc.]

Progress notes:
[Paste notes]

Please produce:
1. Short status summary
2. Completed work
3. Current priorities
4. Blockers or risks
5. Decisions needed
6. Next steps
7. Items that need confirmation before sharing

Rules:
- Do not invent progress, metrics, owners, or dates.
- Mark uncertain items as "Needs confirmation."
- Keep the update clear and concise.

Expected output

A clear status update with progress, priorities, blockers, decisions, next steps, and confirmation items.

Human review point

Before sharing, confirm facts, metrics, risks, client-sensitive wording, and any commitments.

Risk level

Medium if shared with clients, sponsors, or leadership.

Common failure mode

AI makes the project sound healthier or more complete than it is.

How to improve the workflow

Add a required status format, verified metric fields, and a rule to separate facts from interpretation.

How to document it as an SOP

Define update cadence, source notes, reviewer, status format, and where updates are posted.

Workflow 6: Meeting-to-action-items workflow

Who it is for

Small teams, creators, agencies, consultants, founders, and operators turning meetings into next steps.

When to use it

Use this after a planning meeting, client call, sponsor call, editorial meeting, or internal check-in.

Inputs needed

Meeting notes, transcript excerpt, attendees, project context, and known deadlines.

Suggested tools

AI assistant, notes document, task manager, and human meeting owner.

Step-by-step workflow

• Paste meeting notes.
• Ask AI to extract decisions, tasks, owners, deadlines, and open questions.
• Human confirms every action item.
• Move approved items to the task system.
• Send a reviewed summary if needed.

Copy-paste prompt

You are helping me turn meeting notes into reviewed action items.

Meeting context:
[Describe the meeting]

Meeting notes:
[Paste notes]

Please produce:
1. Decisions made
2. Action items
3. Owner for each action item, only if clearly mentioned
4. Deadline, only if clearly mentioned
5. Open questions
6. Items that need confirmation
7. Draft internal recap for review

Rules:
- Do not invent owners, deadlines, or decisions.
- Mark unclear items as "Needs confirmation."
- Keep the output easy to move into a task manager.

Expected output

Decisions, action items, owners/deadlines if known, open questions, confirmation items, and a recap draft.

Human review point

Before assigning tasks or sending the recap, confirm every owner, deadline, and decision.

Risk level

Low internally. Medium if sent to clients or used to assign team commitments.

Common failure mode

AI turns discussion ideas into confirmed tasks.

How to improve the workflow

Add meeting type, task format, and examples of confirmed vs. unconfirmed action items.

How to document it as an SOP

Document who runs the workflow, where notes come from, how tasks are confirmed, and where approved tasks go.

Workflow 7: Internal knowledge base cleanup

Who it is for

Small teams, agencies, creators with internal docs, founders, operators, and support teams.

When to use it

Use this when docs are outdated, duplicated, hard to search, or spread across too many places.

Inputs needed

Doc titles, excerpts, folder list, known outdated docs, owner names if known, and cleanup goal.

Suggested tools

AI assistant, internal wiki, document folder, spreadsheet inventory, and human doc owner.

Step-by-step workflow

• List the docs or paste safe excerpts.
• Ask AI to group, rename, and flag duplicates or outdated docs.
• Human verifies what should be kept, updated, archived, or deleted.
• Update the knowledge base manually.
• Record ownership and review date.

Copy-paste prompt

You are helping me clean up an internal knowledge base.

Cleanup goal:
[Find duplicates, organize docs, update old docs, improve search, etc.]

Document list or excerpts:
[Paste titles, descriptions, or safe excerpts]

Please produce:
1. Suggested doc categories
2. Duplicate or overlapping docs
3. Docs that may be outdated
4. Suggested clearer titles
5. Missing docs or gaps
6. Review questions for the doc owner
7. Cleanup action list

Rules:
- Do not suggest deleting anything without human review.
- Do not invent doc contents.
- Mark uncertain recommendations as "Needs review."
- Keep the cleanup simple.

Expected output

Categories, duplicate candidates, outdated docs, title suggestions, gaps, review questions, and cleanup actions.

Human review point

Before archiving, deleting, renaming, or moving docs, confirm with the doc owner.

Risk level

Medium because bad cleanup can hide or remove useful information.

Common failure mode

AI recommends deleting or merging docs based only on similar titles.

How to improve the workflow

Add owner names, last updated dates, usage notes, and a rule that deletion requires approval.

How to document it as an SOP

Define inventory format, review cadence, archive rules, owner approval, and where cleanup notes are stored.

Workflow 8: Weekly operations review

Who it is for

Founders, solo operators, small teams, agencies, creators, and operations leads.

When to use it

Use this once a week to review what is stuck, what changed, what needs attention, and what can be simplified.

Inputs needed

Task list, project notes, blockers, customer issues, metrics if verified, team notes, and upcoming deadlines.

Suggested tools

AI assistant, project board, weekly review doc, calendar, and human owner.

Step-by-step workflow

• Gather the week’s notes and tasks.
• Ask AI to summarize themes, blockers, risks, and next actions.
• Human checks priorities and commitments.
• Choose a small number of actions for next week.
• Record improvements to the workflow.

Copy-paste prompt

You are helping me run a weekly operations review.

Week:
[Date or week]

Notes, tasks, blockers, and updates:
[Paste notes]

Please produce:
1. What got done
2. What is stuck
3. Top operational risks
4. Decisions needed
5. Follow-ups not to forget
6. Suggested priorities for next week
7. One process improvement to consider

Rules:
- Do not invent metrics, deadlines, or commitments.
- Mark unclear items as "Needs confirmation."
- Keep the review practical for a small team or solo operator.

Expected output

A weekly review with completed work, stuck items, risks, decisions, follow-ups, next priorities, and one improvement idea.

Human review point

Before setting next week’s plan, confirm priorities, deadlines, and whether the suggested improvement is worth doing.

Risk level

Low for personal review. Medium if it changes team priorities or client commitments.

Common failure mode

AI suggests too many priorities and makes the next week more complicated.

How to improve the workflow

Add constraints like available hours, top goal, and maximum number of priorities.

How to document it as an SOP

Define review day, required inputs, output format, reviewer, and where weekly decisions are saved.

Mini exercise

Choose one messy internal process or weekly review. Run one workflow with safe notes. Then write the human review checklist and one SOP rule that would keep the workflow from adding complexity.

Safe Codex workflow for non-coders

1. Start with inspection: ask the agent to read the relevant files and explain what they do.
2. Ask for a plan: request the smallest safe change and the files it expects to edit.
3. Confirm the plan if you understand it. If not, ask the agent to simplify the explanation.
4. Let the agent edit one focused thing.
5. Ask for a plain-English summary of what changed.
6. Preview the page and test the specific behavior that changed.
7. Ask the agent to check for bugs, mobile issues, accessibility issues, and JavaScript errors.
8. Keep rollback notes so you know what to undo if the change causes trouble.

Do not change code yet. First inspect the file and explain what it does.

File or page:
[Paste file name, page name, or current URL]

My goal:
[Describe the improvement]

Please explain:
1. What files or sections seem relevant
2. What the current code appears to do
3. What you would change
4. What could break
5. How we should test the change

I want help describing a small website feature before any code is changed.

Page or file:
[Paste page, file, or section]

Audience:
[Who uses it]

Goal:
[What the user should be able to do]

Constraints:
[WordPress, no external libraries, mobile friendly, brand tone, etc.]

Must not change:
[List anything that should stay the same]

Please turn this into a clear coding-agent task with acceptance criteria.

Make the smallest safe change possible.

Issue:
[Describe the issue]

Rules:
- Do not redesign unrelated sections.
- Do not rename classes unless necessary.
- Do not add external libraries.
- Preserve the existing style and structure where possible.
- Explain any change that could affect other parts of the page.

Before editing, tell me what files you plan to change and why.

Check for anything that could break inside a WordPress Custom HTML block.

Please review:
1. Whether the snippet avoids html, head, and body tags
2. Whether CSS is scoped to this block
3. Whether JavaScript is scoped and safe to run on the page
4. Whether buttons and expandable sections still work
5. Whether anything depends on external libraries, fonts, images, or tracking
6. Whether mobile layout could overflow or overlap

Do not change code yet. Give me the risks first.

Review this for mobile layout issues, broken buttons, and JavaScript errors.

Please test:
1. Desktop layout
2. Mobile layout
3. Any buttons or expandable sections touched by the change
4. Text overflow or overlapping elements
5. Console errors or warnings
6. Whether the change matches the original goal

Report the result in plain English with:
- Passed checks
- Issues found
- Recommended next fix, if any

Before I use this change, give me a rollback checklist.

Include:
1. Files or sections changed
2. What changed in plain English
3. How to undo the change
4. What to check after rolling back
5. Any parts that would be hard to undo

If the change is risky, tell me why.

This code will be pasted into a WordPress Custom HTML block.

Requirements:
- Do not include html, head, or body tags.
- Use one wrapper div.
- Scope all CSS to the wrapper.
- Prefix all custom classes with [prefix].
- Use vanilla JavaScript only.
- Do not use external libraries, external fonts, external images, or tracking scripts.
- Make it mobile responsive.
- Keep buttons accessible.

Before editing, tell me whether the current code follows these rules.

After editing, give me a plain-English summary of what changed.

Include:
1. Files or sections changed
2. What changed for the user
3. Why the change was made
4. How you tested it
5. Anything I should review manually
6. Any remaining risks or limitations

Avoid technical jargon unless you explain it.

Act as a careful reviewer of the change you just made.

Find bugs or risks related to:
1. Broken buttons or links
2. JavaScript errors
3. Mobile layout issues
4. Text overflow or overlapping sections
5. WordPress Custom HTML compatibility
6. Accessibility problems
7. Anything that changed outside the requested scope

Give me:
- Findings ordered by severity
- File or section references
- A small fix plan for each issue
- Anything that needs human review

Review this page for accessibility and mobile layout improvements.

Focus on:
1. Heading hierarchy
2. Button labels and aria-expanded state where needed
3. Keyboard-friendly interactions
4. Color contrast
5. Readable mobile text sizes
6. Tap target spacing
7. Text overflow, overlap, or horizontal scrolling
8. Layout issues inside a WordPress content area

First list issues and recommended fixes. Do not edit until I approve.

WordPress Custom HTML safety checklist

• The snippet does not include html, head, or body tags.
• There is one clear wrapper element for the whole block.
• Custom CSS is scoped to the wrapper.
• Custom class names use a unique prefix.
• JavaScript looks for elements inside the wrapper instead of the whole page when possible.
• No external libraries, fonts, images, or tracking scripts were added without a clear reason.
• Buttons have clear labels and correct expanded or hidden states when used.
• The page works on mobile without horizontal scrolling.
• The snippet still works if WordPress places other content above or below it.
• You previewed it before publishing.

Rollback checklist

• I saved or can access the previous version.
• I know which files or sections changed.
• I know what to undo if the page breaks.
• I checked that no unrelated files were changed.
• I know how to restore the previous WordPress block, draft, or local file.
• I know what to test after rollback.

Testing checklist

• Open the page where the change appears.
• Check the changed section on desktop.
• Check the changed section on mobile.
• Click every changed button, link, accordion, tab, or form control.
• Look for text overflow, overlap, cramped spacing, and horizontal scrolling.
• Check for JavaScript errors or warnings.
• Confirm the change matches the original request.
• Ask for a plain-English summary of tests run and issues found.

Mini exercise

Choose one low-risk change, such as improving button spacing, clarifying a section title, or fixing a small mobile layout issue. Run the safe workflow: inspect first, plan first, make the smallest safe edit, test, ask for a summary, and create rollback notes.

Risk scoring framework

Score each category from 0 to 2. Add the score before deciding how much review the workflow needs.

Score guide: 0 to 3 is usually low risk, 4 to 7 is medium risk, and 8 or higher is high risk. Any legal, medical, financial, employment, customer data, safety, contract, pricing, public claim, or sensitive support workflow should be treated as high risk even if the score looks lower.

Copy-paste prompt pack: identify risks and uncertainties

General risk scan

Identify risks and uncertainties in this AI workflow.

Workflow:
[Describe the workflow]

Input:
[What the AI will receive]

Output:
[What the AI will create]

Action after output:
[Draft only, send, publish, update, decide, delete, recommend, etc.]

Please identify:
1. Risk level: low, medium, or high
2. Who could be affected if the output is wrong
3. Private, customer, legal, financial, medical, employment, or public-claim concerns
4. Assumptions the AI would have to make
5. Uncertainties that should be shown to the human reviewer
6. Human approval checkpoint
7. Whether this should be automated, assisted only, or kept manual

Uncertainty check

Review this AI output for uncertainty before I use it.

AI output:
[Paste output]

Source material:
[Paste safe source material or summarize it]

Please mark:
1. Claims directly supported by the source
2. Claims that are plausible but not proven
3. Claims that need a source
4. Missing context
5. Places where the wording sounds too certain
6. Safer wording
7. What a human must verify before use

I want to check whether this is a low-risk AI workflow.

Task:
[Describe the task]

Input:
[Describe the information I will provide]

Output:
[Describe what AI will create]

Please tell me:
1. Whether this appears low risk, medium risk, or high risk
2. What could still go wrong
3. What I should review manually
4. Whether any private, customer, legal, financial, medical, or public-claim risk is present
5. How to keep this workflow lightweight but safe

This is a medium-risk workflow because the output may affect another person or client.

Task:
[Describe the task]

Audience:
[Who will receive or rely on the output]

Source material:
[Paste safe source material]

Please draft the output, but also include:
1. Assumptions you made
2. Uncertainties or missing context
3. Claims that need human verification
4. Promises, deadlines, prices, or commitments that need approval
5. A human review checklist before use

Treat this as a high-risk workflow.

Task:
[Describe the task]

Why it may be risky:
[Legal, financial, medical, employment, customer data, public claims, sensitive support, pricing, contract, safety, or other concern]

Please do not give final advice or make a final decision.
Instead, help me:
1. Organize the facts I already have
2. List questions a qualified human should answer
3. Identify missing information
4. Identify possible risks and uncertainties
5. Draft a review checklist for the expert or responsible owner

Help me decide whether this workflow should be automated.

Workflow:
[Describe the workflow]

Please answer:
1. What action would the AI take or suggest?
2. Who could be affected if it is wrong?
3. What data would the AI need?
4. Is the output public, external, consequential, or hard to undo?
5. Should this be automated, assisted only, or kept manual?
6. What human approval checkpoint is required?

“Should this be automated?” decision tree

1. Does the workflow use private, customer, employee, medical, financial, legal, or sensitive data? If yes, do not automate final action without approval.
2. Will the output be sent, published, priced, promised, or used to decide something important? If yes, human approval is required.
3. Could the output harm someone if wrong? If yes, treat it as high risk and involve a responsible human or expert.
4. Is the task easy to undo and based on your own non-sensitive notes? If yes, it may be low risk.
5. Can you test the workflow on old examples before using it? If no, keep it manual until you can test.
6. Can you clearly explain the human review point? If no, the workflow is not ready.

Review this workflow for privacy and customer data risk.

Workflow:
[Describe the workflow]

Data involved:
[List the data types]

Tool or tools:
[List tools if known]

Please identify:
1. Private or sensitive data in the workflow
2. Data that can be removed or replaced with placeholders
3. Permissions or access risks
4. Customer data risks
5. Human approval points
6. Whether a simpler, safer version is possible

Review this draft for public claims that need checking.

Draft:
[Paste draft]

Please create a claims table with:
1. Claim
2. Type: fact, opinion, promise, comparison, statistic, or recommendation
3. Source needed
4. Risk if wrong
5. Suggested safer wording
6. Whether it must be verified before publishing

This involves legal, financial, or medical caution.

Topic:
[Describe the topic]

Please do not provide final advice.
Instead, help me prepare for human expert review by creating:
1. A plain-English summary of the issue
2. Questions to ask a qualified professional
3. Missing information
4. Possible risks or tradeoffs
5. Draft language that is clearly marked "Requires expert review"

Help me design human approval checkpoints for this AI workflow.

Workflow:
[Describe workflow]

Output:
[What AI produces]

Possible action after output:
[Send, publish, update, decide, delete, recommend, etc.]

Please define:
1. What action must be blocked until review
2. Who should review it
3. What they must check
4. What counts as approval
5. What should happen if the reviewer is unsure
6. What should be documented

Help me run a quality control loop on this AI output.

Original task:
[Describe task]

Source material:
[Paste or summarize source]

AI output:
[Paste output]

Please review for:
1. Accuracy against the source
2. Missing information
3. Unsupported claims
4. Tone problems
5. Privacy or customer data issues
6. Risk level
7. Specific corrections
8. Prompt improvements for next time

Help me design a test plan for this AI workflow.

Workflow:
[Describe workflow]

Success standard:
[What a good output must do]

Known risks:
[List risks]

Please create:
1. 3 to 5 safe test cases
2. What input each test needs
3. What a good output should include
4. What failure would look like
5. A scoring checklist
6. What to change if the workflow fails

Workflow QA checklist

• The task is clearly defined.
• The trigger is clear.
• The input is safe and necessary.
• The tool is appropriate for the data and task.
• The prompt tells AI not to invent missing facts.
• The output format is easy to review.
• The success standard is specific.
• Failure warning signs are documented.
• The risk score has been assigned.
• The human approval point is named.
• The workflow has been tested on safe examples.
• The SOP explains what to do when the output is uncertain or wrong.

Look for red flags in this AI workflow or output.

Workflow or output:
[Paste workflow or output]

Please identify:
1. Hallucinated or unsupported facts
2. Privacy or customer data issues
3. Public claims that need sources
4. Legal, financial, medical, employment, or sensitive support concerns
5. Actions that are hard to undo
6. Places where AI sounds more certain than the evidence supports
7. Whether we should stop, revise, or ask a human expert

Help me prepare a handoff to a human expert or responsible owner.

Situation:
[Describe situation]

AI output or draft:
[Paste output if safe]

Please create:
1. A plain-English summary
2. The decision that needs human expertise
3. Facts already known
4. Uncertainties or missing information
5. Risks if handled incorrectly
6. Questions for the expert
7. What action should stay blocked until approval

Mini exercise

Choose one workflow you want to use at work. Score it with the risk framework, run the “should this be automated?” decision tree, write the human approval checkpoint, and create a QA checklist. If the workflow scores high risk, rewrite it as an expert handoff workflow instead of an automation.

The beginner stack rule

Start with the fewest tools that let you run the workflow manually: one AI assistant, one place for source material, one place for outputs, one review checklist, and one owner. Add automation only after the manual workflow works.

Help me choose a simple AI workflow stack based on the work, not tool hype.

My role:
[Creator, marketer, founder, student, operator, agency owner, solo business owner, etc.]

Repeatable task:
[Describe the task]

Current tools I already use:
[List tools or categories]

Input:
[What information starts the workflow]

Output:
[What the workflow should produce]

Please recommend:
1. The minimum tool categories I need
2. What each tool category should do
3. What I should not overcomplicate
4. Whether a spreadsheet is enough
5. Whether an automation tool is needed yet
6. Human review point

Map my workflow to tool categories.

Workflow:
[Describe workflow]

Please fill this out:
1. Chat assistant role:
2. Research assistant role:
3. Spreadsheet role:
4. Project manager role:
5. Automation tool role:
6. Document hub role:
7. Optional coding agent role:

For any category I do not need yet, say "skip for now" and explain why.

Tool selection checklist

• What task will this tool help with?
• What input does it need?
• What output should it create?
• Who reviews the output?
• Where will the approved output be stored?
• Does the tool need access to private or customer data?
• Can the workflow run without this tool?
• Does this tool reduce manual confusion or add a new place to check?
• Can I explain the tool’s role in one sentence?
• What would I use if this tool disappeared tomorrow?

Do I really need an automation tool for this workflow yet?

Workflow:
[Describe workflow]

How often it happens:
[Daily, weekly, monthly, occasional]

Current manual steps:
[List steps]

Problems with the manual version:
[List problems]

Please tell me:
1. Whether I should automate now, later, or not at all
2. What must be proven manually first
3. What should stay manual
4. What human approval point is required
5. What could break if I automate too early

Can this workflow be managed with a spreadsheet first?

Workflow:
[Describe workflow]

Items to track:
[Leads, content ideas, tasks, competitors, customer themes, etc.]

Please suggest:
1. Spreadsheet columns
2. Status labels
3. Review fields
4. Risk fields
5. What should not be automated yet
6. When I would outgrow the spreadsheet

Does this workflow need a coding agent?

Workflow:
[Describe workflow]

Technical artifact involved:
[Website page, Custom HTML block, script, internal tool, none, etc.]

Please tell me:
1. Whether a coding agent is useful here
2. The smallest safe technical task it should do
3. What it should inspect before editing
4. What tests are required
5. What rollback note is required
6. When I should ask a developer to review

Example stack 1: Creator

Main job to be done

Turn one video, interview, livestream, or research session into reusable content ideas and publishing assets.

Recommended type of tool

Chat assistant for repurposing, document hub for transcripts and prompts, spreadsheet for content inventory, optional research assistant for source checks, project manager if multiple posts need tracking.

Why it matters

Creators need consistency without flooding channels with generic output.

What not to overcomplicate

Do not add automation before your content review checklist catches generic wording, missing first-hand perspective, and unsupported claims.

First workflow to build

Transcript to newsletter outline, short-form angles, and thumbnail concept list.

Human review point

Review for accuracy, voice, original opinion, and whether the content reflects real experience.

Example stack 2: Marketer

Main job to be done

Turn campaign notes, customer language, and performance observations into a clearer next campaign plan.

Recommended type of tool

Chat assistant for positioning drafts, research assistant for claim and competitor checks, spreadsheet for campaign ideas and scores, document hub for briefs, project manager for launch tasks.

Why it matters

Marketing workflows need traceable claims, clear audience assumptions, and review before anything goes public.

What not to overcomplicate

Do not automate publishing or outreach until claims, audience fit, and approval rules are stable.

First workflow to build

Landing page teardown to improvement backlog with claim checks and priority scores.

Human review point

Review public claims, audience assumptions, tone, and whether recommendations match actual campaign data.

Example stack 3: Founder

Main job to be done

Turn scattered market notes, customer calls, and product ideas into decisions and next actions.

Recommended type of tool

Chat assistant for synthesis, document hub for customer notes and decision logs, spreadsheet for lead or feedback tracking, project manager for execution, research assistant for source checking.

Why it matters

Founders need faster thinking without losing the source of the evidence.

What not to overcomplicate

Do not connect every tool in the business. Start with one decision workflow and one storage location for evidence.

First workflow to build

Customer call notes to pain-point themes, open questions, and product follow-up tasks.

Human review point

Review whether AI overgeneralized from too little evidence or turned opinions into facts.

Example stack 4: Student

Main job to be done

Turn class notes, readings, and assignment requirements into study plans, summaries, and draft outlines while preserving learning.

Recommended type of tool

Chat assistant for explanations and practice questions, document hub for notes, spreadsheet for study tracker, research assistant for source checking when allowed by the assignment.

Why it matters

The stack should help the student understand the material, not outsource the learning or violate academic rules.

What not to overcomplicate

Do not build automations for assignments. Keep the focus on comprehension, planning, and review.

First workflow to build

Lecture notes to summary, unclear concepts list, practice questions, and a study schedule.

Human review point

Check course rules, verify facts against class material, and make sure the final work is genuinely your own.

Example stack 5: Operator

Main job to be done

Turn messy internal notes, tasks, and status updates into clear priorities, owners, and follow-ups.

Recommended type of tool

Chat assistant for summarizing and triage, project manager for owners and dates, document hub for SOPs, spreadsheet for recurring review logs, optional automation tool after the process is stable.

Why it matters

Operations workflows should make work clearer, not create another system people forget to update.

What not to overcomplicate

Do not automate task creation until owners, deadlines, and review rules are consistent.

First workflow to build

Weekly operations review to blockers, decisions, follow-ups, and next-week priorities.

Human review point

Review owners, deadlines, commitments, and whether the plan is realistic.

Example stack 6: Agency owner

Main job to be done

Turn client inputs into briefs, tasks, drafts, review notes, and delivery checklists.

Recommended type of tool

Chat assistant for draft briefs and summaries, document hub for client context, project manager for deliverables, spreadsheet for pipeline and QA tracking, research assistant for claim checks.

Why it matters

Agency workflows need consistency, but clients still expect judgment, confidentiality, and review.

What not to overcomplicate

Do not use automation to send client messages, publish work, or update client-facing assets without approval.

First workflow to build

Client intake notes to project brief, open questions, task list, and review checklist.

Human review point

Review client promises, scope, tone, confidential details, and anything that affects delivery expectations.

Example stack 7: Non-technical solo business owner

Main job to be done

Turn recurring admin, marketing, customer questions, and planning notes into clearer drafts and checklists without building a complex system.

Recommended type of tool

Chat assistant for drafts and planning, document hub for reusable prompts and SOPs, spreadsheet for simple tracking, project manager only if tasks are being missed, optional coding agent only for small website snippet help.

Why it matters

Solo operators need tools that reduce mental load and are easy to maintain.

What not to overcomplicate

Do not connect tools just because they can connect. If you cannot explain the workflow on one page, simplify it.

First workflow to build

Weekly business planning assistant that reviews notes, open tasks, customer questions, and top priorities.

Human review point

Review customer-facing wording, promises, prices, and anything involving private customer information.

Mini exercise

Choose one repeatable task from your work. Build a minimum stack with no more than four categories. Run the workflow manually once, then write down what tool you were tempted to add and why you are waiting.

What you are building

One reusable AI workflow for a real work task. It should be narrow enough to test and useful enough to repeat.

• Good project: “Turn meeting notes into a client-safe summary and action list.”
• Good project: “Turn sponsor lead notes into a reviewed outreach draft.”
• Good project: “Turn a weekly task dump into a prioritized planning brief.”
• Too broad: “Automate my marketing.”
• Too risky for a beginner: “Let AI respond to customers automatically.”

Help me choose one final project workflow.

My role:
[Creator, marketer, founder, student, operator, agency owner, solo business owner, etc.]

Tasks I repeat:
[List 3 to 5 messy repeatable tasks]

Please recommend the best first workflow based on:
1. Usefulness
2. How easy it is to review
3. Risk level
4. Whether the input is safe to use
5. Whether the workflow can become an SOP

Do not recommend high-risk automation for my first project.

Help me rewrite this task clearly.

Messy task:
[Describe the task in normal language]

Please turn it into this format:
Turn [input] into [output] so [person] can [next action].

Also list:
1. What is included
2. What is not included
3. What the workflow should not do
4. What human review is needed

Help me define the input for this workflow.

Workflow task:
[Paste clear task sentence]

Possible input:
[Describe the source material]

Please identify:
1. Required input
2. Optional input
3. Input that should be removed for privacy
4. Input that may need permission
5. What the AI should do if input is missing or unclear

Help me choose the simplest tool stack for my final project.

Workflow task:
[Paste task]

Input:
[Describe input]

Output:
[Describe expected output]

Current tools I already use:
[List tools or categories]

Please recommend:
1. The minimum tool categories needed
2. What each tool does
3. What I should skip for now
4. Whether a spreadsheet is enough
5. Whether automation should wait
6. Human review point

You are helping me run a reviewable AI workflow.

Workflow task:
[Turn input into output so person can next action]

Input:
[Paste source material]

Rules:
- Use only the provided input unless I explicitly ask for outside research.
- Do not invent facts, dates, names, numbers, quotes, sources, promises, or decisions.
- Mark missing or unclear information as "Needs confirmation."
- Keep the output practical and easy for a human to review.
- Do not send, publish, or take action.

Please produce:
1. [Output section 1]
2. [Output section 2]
3. [Output section 3]
4. Assumptions and uncertainties
5. Human review checklist
6. Suggested next action

Help me define the expected output for this workflow.

Workflow task:
[Paste task]

Who will use the output:
[Person or role]

Next action after output:
[Review, send, publish, decide, plan, update, etc.]

Please create:
1. Output sections
2. Recommended length
3. Formatting rules
4. Success standard
5. Failure warning signs
6. Human review checklist

Define the human review point for this workflow.

Workflow:
[Describe workflow]

AI output:
[Describe output]

Possible next action:
[Send, publish, decide, update, create tasks, etc.]

Please define:
1. What action is blocked until review
2. Who reviews it
3. What they must check
4. What counts as approval
5. What to do if the reviewer is unsure

Score the risk of my final project workflow.

Workflow:
[Describe workflow]

Input:
[Describe input]

Output:
[Describe output]

Next action:
[Describe what happens after the output]

Please score each category from 0 to 2:
1. Impact
2. Data sensitivity
3. Public visibility
4. Reversibility
5. Confidence
6. Automation level

Then tell me:
1. Total risk score
2. Risk level: low, medium, or high
3. Required human review point
4. What to change if this is too risky for a first project

Help me test this AI workflow.

Workflow:
[Describe workflow]

Prompt:
[Paste prompt]

Expected output:
[Describe expected output]

Test output:
[Paste AI output from a test run]

Please evaluate:
1. Did it follow the prompt?
2. Did it use the input accurately?
3. Did it invent anything?
4. Did it mark uncertainty?
5. Did it match the expected output format?
6. What should be improved in the prompt, input, or review checklist?

Help me improve this workflow after testing.

Original workflow:
[Describe workflow]

What failed or felt weak:
[Describe issue]

Current prompt:
[Paste prompt]

Please suggest:
1. One improvement to the input
2. One improvement to the prompt
3. One improvement to the expected output
4. One improvement to the human review checklist
5. One thing that should stay manual
6. A revised version of the prompt

Turn my tested AI workflow into a reusable SOP.

Workflow name:
[Name]

Task:
[Clear task]

Input:
[Input]

Tool:
[Tool or tool category]

Prompt:
[Paste prompt]

Expected output:
[Output]

Risk score and review point:
[Risk and review]

Testing notes:
[What happened during testing]

Please write a simple SOP with:
1. Purpose
2. When to use this workflow
3. Required input
4. Step-by-step instructions
5. Prompt
6. Human review checklist
7. Success standard
8. Failure warning signs
9. What to do when the output is wrong
10. Owner and update frequency

Filled-in example: meeting notes to follow-up summary

Blank final project worksheet

Grading and self-assessment rubric