The AI Prompt Vault: 60+ Copy-Paste Prompts

This section helps you find out what your market actually thinks, what words they use, and where your messaging is broken — before you spend a pound on distribution.

Homepage copy audit

Run this before any paid campaign. If Claude can't identify your customer from your homepage, neither can the people landing on it.

Read this homepage copy. Tell me exactly what product/service this company sells, who it's for, and what makes it different from competitors - using ONLY the words on this page. Then tell me what's missing.

Paste your homepage copy beneath the prompt.

Voice of customer mining

Pull 10–15 real customer reviews from G2, Trustpilot, Google, or the App Store and paste them all in. The phrases Claude pulls out go verbatim into your ads and landing pages.

Analyse these customer reviews. Extract: (1) the top 3 emotional pain points before they found us, (2) the exact phrases they use to describe success, and (3) any words that appear 3+ times. Format as a table.

Proprietary framework builder

Run before writing any long-form content. Naming your methodology is what makes content shareable — generic advice gets ignored, named frameworks get cited.

I'm writing a blog post for [target audience] about [topic]. Before I write it, help me develop a proprietary framework or model that organises this topic in a way no competitor has named yet. Give me 3 options with a name, a 1-line description, and a 3-step structure for each.

Brand voice bible

One-time setup. Paste in 5 of your best-performing pieces of content, then save the output and share it with every writer, agency, and AI prompt from that point forward.

Act as a brand strategist. I'm going to give you 5 pieces of our best-performing content. Analyse them and produce a Brand Voice Guide with: (1) 3 words that describe our tone, (2) 3 words that describe what our tone is NOT, (3) sample sentences showing the right vs. wrong way to write for us, (4) our vocabulary preferences (words we use, words we avoid), and (5) our signature structural patterns.

Competitor ad deconstruction

Go to the Meta Ad Library and find 3–5 competitor ads that have been running for 90+ days — long-running means it's working. Copy the ad text and paste it in.

Deconstruct these ads. For each one: (1) identify the core emotional trigger being used, (2) name the copywriting formula (e.g. PAS, AIDA, before/after), (3) identify what fear or desire is being activated, (4) suggest how we could ethically counter-position against each one without naming the competitor.

RITE prompt framework

Use this structure for any AI task that matters: Role + Input + Task + Example. The more specific each section, the better the output on the first try.

ROLE:
Act as a [specific profession] with [X years experience] in [specific domain]. You are [key traits] and skilled at [specific capabilities].

INPUT:
[Your business/situation]
[Target audience details]
[What you have/don't have]
[What you need]

TASK:
Create a [type of deliverable] with these sections:
1. [Section name] ([length])
   - [Specific requirements]
2. [Section name] ([requirements])

Tone: [Specific tone]
Format: [Output format]
Length: [Total length]

EXAMPLE:
Here's the style and depth I want:
[Good example]
[Bad example]
See the difference? The good version is:
- [Quality 1]
- [Quality 2]

Devil's advocate campaign review

Run before any campaign goes live. Claude will find the thing your team was too excited to see.

Here is our upcoming campaign strategy: [paste strategy]. Play the role of a cynical, experienced marketing critic who has seen 1,000 campaigns fail. Find every possible weak point - messaging gaps, audience assumptions, competitive blindspots, and execution risks. Be brutal. Don't soften anything.

ICP research document

Run before writing any messaging, positioning, or content strategy. Gives you the full customer picture in one structured document.

ROLE:
Act as a customer research specialist with 10 years of experience in B2B SaaS. You're analytical, methodical, and skilled at uncovering pain points, behavioral patterns, and buying triggers.

INPUT:
[Your business/situation]
[Target audience: company type, size, role]

TASK:
Create a comprehensive ICP research document with these sections:

1. ICP Profile Overview (200 words)
   - Demographics, firmographics, psychographics
   - Key characteristics and identifiers

2. Pain Points and Challenges (5-7 bullet points)
   - Ranked by urgency/severity
   - Include emotional and practical dimensions

3. Buying Triggers (3-5 scenarios)
   - What makes them start searching for a solution?
   - What pushes them from thinking about it to ready to buy?

4. Objections and Barriers (4-6 items)
   - Why they wouldn't buy
   - What holds them back

5. Research Validation Questions (10 questions)
   - Questions to ask in discovery calls to validate these assumptions

Tone: Analytical but not academic. Direct, actionable.
Format: Markdown with clear headers and bullet points.
Length: 1000-1200 words.

LinkedIn post framework

Write a LinkedIn post about [topic/story] using this structure: (1) a one-sentence hook that creates a knowledge gap or makes a bold claim, (2) 4-5 short paragraphs that deliver the insight using a real story or specific numbers, (3) a closing line that asks ONE specific, easy-to-answer question. Tone: direct, slightly provocative, no corporate language. Write 3 variations.

Always ask for 3 variations. The first is usually the safest. The third is usually the best.

Content repurposing: 4 formats from one article

Paste any 2,000-word article into Claude and run these four prompts back-to-back. One article, four formats, under 15 minutes.

Extract 7 standalone insights from this article that each work as a single tweet. No fluff, no "thread incoming."

Turn the main argument of this article into a LinkedIn carousel outline - 8 slides, each with a 5-word title and 2-line description.

Rewrite the core lesson of this article as a 60-second YouTube Shorts script. Hook - Problem - Insight - CTA.

Pull out 3 quotes from this article that would work as standalone Instagram caption starters.

LinkedIn engagement machine: 4-prompt chain

Run all four prompts in the same conversation thread. Each step builds on the previous one — the AI retains your transcript context, so the final post feels authentic rather than generic.

Prompt 1: ideation engine

You are Justin Welsh, the solopreneur who built a $5M business through strategic LinkedIn content. Your job is to analyze content and give me 3-5 killer post ideas.

Your content philosophy:
- Every post solves ONE specific problem
- Start with personal experience, end with actionable value
- Make readers feel seen
- Give away your best stuff for free

I'm going to share [transcript/notes/raw ideas] with you. Extract 3-5 LinkedIn post ideas that will make people stop scrolling.

For each idea:
- Hook Angle: the scroll-stopping opener
- Core Value: the one transformation
- Proof Points: what makes this credible (specific numbers, story, unexpected detail)
- Lead Magnet Hook: what they unlock by commenting

[PASTE TRANSCRIPT/NOTES HERE]

Prompt 2: writing system

You are Justin Welsh, master of LinkedIn content that drives engagement.

Take the POST CONCEPT I've selected and write the complete LinkedIn post ready to publish.

Writing rules:
- Hook: first 3 lines, start with tension or bold statement, specific numbers if possible
- Story: 3-6 short paragraphs max 3 lines each, pull from the transcript for authenticity
- Value: framework as bullets with proof in parentheses - e.g. "Voice-to-text everywhere (I haven't typed an email in 3 months)"
- Reframe: 1-3 lines shifting perspective before CTA
- CTA: clear action, name the deliverable, mild FOMO

Write the post for: [PASTE CHOSEN CONCEPT]

Prompt 3: polish filter

You are Justin Welsh, editing with ruthless precision.

Score this post out of 10. List exactly what needs fixing. Provide the fully edited version. Explain the 3 biggest changes and why they work.

Red flags to remove:
- "I think" / "I believe" (just state it)
- Corporate jargon
- Lists without proof points
- Vague time references ("recently" vs "last Tuesday")
- Weak verbs ("utilize" vs "use")

Now edit this post:
[PASTE DRAFT POST HERE]

Prompt 4: lead magnet builder

Write a detailed, structured resource outline for the lead magnet promised in this post.

Lead magnet idea: [INSERT IDEA]

The lead magnet should:
- Deliver on the exact promise made in the LinkedIn post
- Be immediately actionable (usable today)
- Position me as helpful and strategic

Structure:
1. Cover section (title + one-line promise)
2. Introduction (why this matters, what they'll get)
3. Main content sections (3-5 sections with actionable steps)
4. Quick-start action (one thing to do right now)
5. What's next (how to go deeper)

Write in my exact voice - short sentences, no fluff, direct.

Transcript to content angles

You are Gary Vaynerchuk. You know that 90% of any transcript is noise. Scan for: Stories and Moments (something specific with a turning point), Contradictions and Reframes (conventional wisdom being challenged), Specific Proof Points (numbers, results, timelines), Quotable Lines (short punchy phrases that stand alone).

From everything you flag, pick the 5 strongest. For each:
- ANGLE: one sharp sentence
- Type: Story / Contradiction / Proof Point / Quotable
- Why it hits: 2-3 sentences on tension and scroll-stop value
- Source moment: what was said, paraphrased

[PASTE TRANSCRIPT]

LinkedIn carousel from any content

Step 1 — extract angles

Act as an expert Content Strategist. Analyse this content and give me:

1. Core Thesis (one sentence)
2. Key Data/Facts (specific numbers or case studies)
3. Golden Quotes (3-5 punchy verbatim quotes)
4. The Framework (if there's a step-by-step process)

Then pitch 5 narrow, specific carousel angles. Each angle: catchy headline + one-sentence premise.

[PASTE CONTENT OR URL]

Step 2 — generate the carousel outline

You are a LinkedIn Ghostwriter. Turn this angle into a 10-slide carousel outline:

Angle: [YOUR CHOSEN ANGLE]

Slide breakdown:
- Slide 1: Hook (polarising or curiosity-gap headline)
- Slide 2: Stakes (why this matters, cost of inaction)
- Slide 3: Core insight or golden quote
- Slides 4-8: One aha moment per slide (specific data/steps)
- Slide 9: TL;DR checklist
- Slide 10: CTA question to spark comments

Format each as:
[Slide #] [Headline]
[Text content]

Then take the outline to Gamma.app: Create → Studio mode → Social format → 10 cards → Portrait → "Just vibes" text quantity. Paste the outline.

Deep research for content strategy

You are Derral Eves, YouTube strategist with deep expertise in analysing what makes content perform. Use deep research to build a comprehensive content intelligence report on this topic.

Topic: [INSERT TOPIC]
Channel context: [YOUR CHANNEL/BRAND DESCRIPTION]

Cover:
1. Top Performing Content - find strongest pieces on this topic. What formats, hooks, and angles keep appearing at the top?
2. Written Content Landscape - how are the top articles structuring information?
3. Audience Sentiment - what questions keep coming up in comments and Reddit? Pull recurring language.
4. What Top Performers Do Well - 3-5 consistent strengths
5. Value-Add Opportunities - where is there room for something the top performers haven't done?
6. Research Summary - 2-3 most promising angles for content that stands out

Email imperfections: make AI emails feel human

Use whenever AI-written emails feel too smooth. Authenticity is now the scarcest resource in email marketing.

Write an email about [topic/offer] for [audience]. After you write it, deliberately add 3 imperfections: one sentence fragment, one place where the writer changes their mind mid-thought, and one moment of genuine vulnerability or uncertainty. The email should feel like it was written at 11pm by a real human who cares.

Subject line battle test

Write 10 subject lines for this email. Include: 2 curiosity-gap lines, 2 number-led lines, 2 that use the word "you", 2 contrarian lines, and 2 ultra-short lines (under 5 words). Then score each one out of 10 for open rate potential and explain why.

Pick your top 3 and A/B test. Claude's scoring won't always be right, but it forces you to think about mechanism, not just words.

5-angle ad brief

Run before writing any ad. Five angles, you pick the best 2, test them, scale the winner.

Our product is [X]. Our target customer is [Y]. Generate 5 completely different creative angles for a Facebook/Instagram ad campaign. Each angle should target a different emotional trigger (fear of missing out, pride, frustration relief, aspiration, belonging). For each angle, give me: the core message, a sample headline, and a 2-line body copy.

PR data study: prompt chain

This chain builds a data study with 500–1,000+ data points that journalists can cite, an unbiased findings report, a press release ready to send, a public webpage where journalists can find and link to your study, and charts they can drop into articles.

Step 3 is the data collection run itself — you set it up in Step 2, so there's no prompt for it.

Step 1: find your question

You are Brian Dean, creator of How to Get Press Every Month, known for engineering data-driven studies that earn authoritative backlinks and land features in major publications.

INPUT
Your business or niche: [DESCRIBE YOUR BUSINESS OR NICHE]
Your target audience: [DESCRIBE WHO YOU SERVE]
Dream publications you want to be featured in: [LIST 3-5 PUBLICATIONS]

TASK
Brainstorm 10 potential data study topics. For each idea:
- Study title as a working headline
- One-sentence description of what would be measured and how
- Rate against: IS IT TRENDING? DOES IT LACK DATA? IS IT TANGENTIAL? (Strong/Medium/Weak each)

Then recommend your top 3 with a paragraph on pickup potential for the listed publications.

Step 2: choose your data source

I've chosen my study topic: [YOUR QUESTION FROM STEP 1]

I need to collect data to answer this question. I need at least 500 data points, ideally 1,000+.

I'm not technical. What are my realistic options for collecting this data with Claude Cowork using connectors like Apify and Firecrawl?

For each option: data source, how to collect it, roughly how long, whether it's free or what it costs, how credible it would look in a press release.

Recommend the best option for credibility and ease.

Step 4: analyse with AI

You are Nate Silver, data journalist and founder of FiveThirtyEight, known for turning raw datasets into compelling narratives that non-technical audiences can immediately understand and quote.

INPUT
Research question: [WHAT ARE YOU TRYING TO FIND OUT?]
[PASTE DATA]

TASK
1. The Headline Number - the single most surprising finding. State as a clean, quotable statistic.
2. Category Breakdowns - break the data down by every available column. Give the key number and one sentence of plain-English interpretation.
3. Quotable Statistics - 10-12 statistics in journalist-ready language. Format: "X% of [group] do/are/have [thing]"
4. Interesting Comparisons - most meaningful contrasts in the data
5. Surprises and Pattern Breaks - anything that goes against conventional wisdom
6. Confidence and Caveats - flag findings that could be misleading

Output as a numbered list, ordered from most interesting to least. Plain English. Write as if briefing a journalist who needs to file in an hour.

Step 5: verify findings

I need you to be my quality control editor. Try to break my findings:

[PASTE YOUR FINDINGS FROM STEP 4]

For each finding:
1. COULD THIS BE A FALSE POSITIVE?
2. IS THE SAMPLE BIASED?
3. IS THE LANGUAGE MISLEADING?
4. WHAT WOULD A SCEPTICAL JOURNALIST ASK?

Be brutal. I'd rather find the problems now than after it's published.

Step 6: write the findings report

Write an unbiased, factual findings report for my study. This is NOT a press release. It should read like a research summary.

Study: [YOUR QUESTION]
Sample: [NUMBER] [DATA TYPE] collected from [SOURCE] between [DATES]
Key findings: [PASTE VERIFIED FINDINGS]

Structure:
1. Headline Findings
2. Detailed Breakdowns
3. Methodology
4. Raw Data Reference

Rules: lead with numbers, not opinions. No adjectives that interpret the data. Every claim needs the exact number AND base numbers ("83 out of 1,019" not just "8.1%"). Sound like a BBC news report.

Step 7: build the press pack

Using only the findings from our study, write a press release:

Headline: Lead with the single most surprising statistic, followed by "according to new study" and the researcher's name. Under 20 words.
Subheadline: Sample size, scope, one supporting statistic.
Lead paragraph: Restate headline stat. Include study link. Add one more key finding. 3-4 sentences max.
Methodology: 2-3 plain-English sentences.
Key Findings: 5 bullet points. Bold the topic label, follow with the result. Precise numbers throughout.
About section: 2-3 sentences. Factual.
Media contact: Name, email, website.

Write 3 email pitches for 3 different publication types. Same data, different angle.

PITCH 1 - [TRADE/INDUSTRY PRESS]
PITCH 2 - [BUSINESS PRESS: Forbes, Fast Company, Business Insider]
PITCH 3 - [NATIONAL PRESS: BBC, Guardian, Times]

Each pitch: under 150 words. Lead with the most relevant stat for THAT audience. Sound human, not corporate.

Step 8: public page

Create a single-page HTML webpage to present my study findings publicly.

Study title: [YOUR TITLE]
Conducted by: [YOUR NAME / COMPANY]
Key findings: [PASTE YOUR FINDINGS]

Include: clear headline with the main stat, key findings summary (8-10 stats), methodology section, note that raw data is available on request, professional clean design, mobile responsive, no email gate.

A journalist landing on this page should immediately think: legitimate research I can cite.

Seven self-refreshing dashboards for Claude Cowork, each pulling live data from your tools so you stop tab-hopping every morning. Build each one by pasting the prompt into a Cowork session with the relevant connectors active.

1. Competitor move tracker

Connectors: none required — Claude browses the web. Stronger with your connectors active.

Build a live Cowork artifact called "Competitor Move Tracker" - a weekly intel report on what my competitors are doing.

My business context: [describe your business in 1-2 sentences]

Competitors to track:
1. [Competitor 1] - [blog URL] - [LinkedIn URL]
2. [Competitor 2] - [blog URL] - [LinkedIn URL]
3. [Competitor 3] - [blog URL] - [LinkedIn URL]

What to show for each competitor:
- Latest blog/news post (title, date, 1-line summary)
- LinkedIn activity in the last 7 days
- Hiring signals - flag with 🚨 if they posted job ads in the last 14 days
- Expansion signals (new offices, partnerships, funding, launches)
- A 1-line "What this means for me" insight based on my business context

Top of page: alert bar showing "This Week's 3 Biggest Moves" across all competitors.

Layout: 2-column grid of competitor cards, each collapsible. Red/amber/green activity dot on each card.
Visual: Intel-report energy. Clean, premium. Dark mode palette.

2. Daily command centre

Connectors: Gmail (or Outlook) + Google Calendar

Build a live Cowork artifact called "Daily Command Centre" - a single-page morning dashboard.

About me: [Your name + 1-line about what you do]

What to show:
1. Today's date - big and clear at the top
2. Today's Calendar - events with time, title, attendees. Highlight back-to-back blocks.
3. Inbox: Needs a Reply - pull unread emails from last 24 hours. Surface only ones that need a reply (skip newsletters and automated emails). Show sender + 1-line summary.
4. Top 3 Priorities - pull from calendar and inbox, suggest 3 things that should get focused time today.
5. Energy check - a small card with a 1-10 slider.

Connectors: Gmail (list_messages, last 24 hours, unread) + Google Calendar (list_events, today)
Layout: Single column, 5 stacked cards
Visual: Warm, calm morning energy. Mobile-friendly width.

3. Daily financial position

Connectors: QuickBooks, Xero, FreeAgent, or Wave

Build a live Cowork artifact called "Daily Financial Position" - a daily snapshot of where my business stands.

My setup:
- Accounting tool: [QuickBooks / Xero / FreeAgent / other]
- Currency: [GBP / USD / EUR]
- Business type: [sole trader / limited company / LLC]

What to show (6 cards in a 3x2 grid):
1. Cash in bank - today's balance + 7-day sparkline
2. This month's revenue - vs last month at same day-of-month, with % change
3. Outstanding invoices - total value + count, flag any overdue
4. Bills due in next 14 days - total value + count
5. Profit this month so far
6. A "What I'd do today" nudge - generated by Claude based on the numbers (e.g. "Chase the 3 invoices over 30 days late")

Layout: 6-card grid (3 columns x 2 rows desktop, single column mobile). Each card: big number, label, % change. Cash card has inline sparkline (Chart.js). Full-width "What I'd do today" panel at bottom.
Visual: Confident, clean. Green for positive, amber only for warnings.

4. SEO pulse

Connectors: Google Search Console MCP + Google Analytics 4 MCP

Build a live Cowork artifact called "SEO Pulse" - a daily snapshot of my site's search performance.

My setup:
- Domain: [yourdomain.com]
- Date range: last 28 days vs previous 28 days
- Main conversion event in GA4: [e.g. newsletter_signup]

What to show:
- Header: 4 key numbers - clicks, impressions, average position, CTR - each with % change
- Top 10 queries driving clicks (query, clicks, position)
- Top 10 landing pages (URL, clicks, conversions)
- Movers and shakers - pages that moved up or down 3+ positions in the last 7 days
- Traffic source breakdown (organic / direct / referral / social / email) as a donut chart
- A "What to do next" line generated by Claude based on what changed

Connectors: Google Search Console MCP + Google Analytics 4 MCP
Layout: Header strip, 2-column body, movers section, donut at bottom right, full-width action panel
Visual: Clean, data-dense. Use Chart.js for the donut.

5. Support pulse

Connectors: Gmail / Outlook + optional CRM (HubSpot / Intercom / Zendesk)

Build a live Cowork artifact called "Support Pulse" - shows what's in the support queue and what needs leadership attention right now.

My setup:
- Support inbox: [support@yourdomain.com or Gmail label]
- CRM: [HubSpot / Intercom / Zendesk / none]
- "Needs my eyes" criteria: refund requests, churn signals, legal mentions, frustrated language, anything sitting 48+ hours

What to show:
1. Needs My Eyes panel - anything flagged as urgent, with sender + 1-line summary + how long waiting
2. Open tickets strip: today, this week, total open, average response time
3. Tickets by priority (P1/P2/P3)
4. Top 5 themes this week - Claude clusters messages into themes (billing issues, feature requests, onboarding confusion, etc.)
5. At-risk customers - anyone showing churn signals

Layout: Red alert bar at top (only red use case), 4-card metric strip, 2-column body, at-risk list at bottom
Visual: Calm under pressure. Red sparingly, only for genuine alerts.

6. Sales pulse

Connectors: CRM (Apollo, HubSpot, Pipedrive, Notion CRM, or Airtable) + Calendar + optional Stripe

Build a live Cowork artifact called "Sales Pulse" - tracks the metrics that matter for my offer.

My setup:
- CRM: [Apollo / HubSpot / Pipedrive / Notion CRM / Airtable]
- Primary offer: [e.g. "1:1 strategy session at £400" or "SaaS subscription at $99/mo"]
- My funnel stages: [list yours, e.g. "Lead - Qualified - Booked - Completed - Paid"]

4 big-number cards at the top:
- New leads this month (vs same point last month)
- [Calls booked / demos scheduled / trials started]
- [Calls completed / demos run / trials converted]
- Revenue this month (vs last month)

Underneath:
- Pipeline funnel visual showing count at each stage
- "Deals to close this week" list: name, opportunity type, days since last activity (colour-coded), value

Layout: 4 metric cards, funnel visual, deals table
Visual: High-energy, motivating. Green for wins, amber for deals going cold (5+ days since activity).

7. YouTube morning brief

Connectors: YouTube Analytics MCP (via YouTube Data API v3). Ask Claude to walk you through building a custom MCP if you haven't set one up.

Build a live Cowork artifact called "YouTube Morning Brief" - a daily snapshot of my channel performance.

My setup:
- Channel ID: [your YouTube channel ID]
- Date range: last 28 days vs previous 28 days, plus a 7-day sparkline trend
- Currency: [GBP / USD]

Header - 5 metric cards:
- Views (last 28 days vs previous, % change)
- Watch time in hours
- Subscribers gained/lost (net)
- Average CTR
- Average view duration

Top 5 performing videos this month:
- Title, views, watch time, CTR, average view duration
- Green dot if CTR above channel average, amber if below

7-day performance trend chart - rolling view count and watch time (Chart.js line chart, two y-axes)

"What to focus on today" nudge - generated by Claude based on the numbers

Layout: 5-card strip, top 5 videos table, 7-day trend chart, full-width action panel
Visual: Clean and energising. Green for gains, red for losses. Dark mode palette.

Client intelligence brief

Run before every client discovery call — four hours of manual research in ten minutes. Swap in the client details and run with Research mode on.

You are a senior business intelligence researcher. Conduct comprehensive due diligence research on a new client.

CLIENT INFORMATION:
- Business Name: [CLIENT NAME]
- Website: [CLIENT WEBSITE]
- Location: [LOCATION]
- Industry/Sector: [INDUSTRY]
- Initial Notes: [ANY NOTES FROM FIRST CONTACT]

RESEARCH ACROSS THESE DOMAINS:

1. Business Profile and Operations
Ownership structure, service/product offerings, team size, locations, technology adoption, reputation signals.

2. Market Positioning and Competitive Landscape
Value propositions, pricing positioning, target customers, 3-5 direct competitors, competitive advantages/disadvantages, underserved segments.

3. Digital Presence and Marketing
Website quality, SEO and local search visibility, social media presence, review profiles with ratings/sentiment, content marketing, paid advertising activity.

4. Financial Indicators and Business Health
Companies House filings (if UK), revenue model, signs of financial health or stress, recent investments.

5. Growth Trajectory and Challenges
Recent developments, recruitment patterns, customer volume trends, regulatory compliance, operational bottlenecks.

6. Industry Context and Opportunities
Local demographic and economic trends, emerging opportunities, regulatory changes, technology trends.

OUTPUT FORMAT:
1. Executive Summary (2-3 paragraphs)
2. Business Profile
3. Market Position Analysis
4. Digital and Marketing Assessment
5. Financial Health Indicators
6. Growth Opportunities and Challenges
7. Key Risks and Considerations
8. Recommended Focus Areas
9. Questions for Client Discovery Meeting
10. Source References

REQUIREMENTS: flag assumptions vs verified facts. Focus on actionable insights. Cite sources for key claims.

AI audit questionnaire

Generates a tailored questionnaire before an AI audit with any business. The output adapts to the specific company type.

You are an experienced AI consultant and operational auditor specialising in identifying inefficiencies and automation opportunities.

Business type: [INSERT BUSINESS TYPE]
Industry: [INSERT INDUSTRY]
Team size: [INSERT TEAM SIZE]

Generate a tailored AI audit questionnaire for this specific business. Organise into relevant sections only (operations, marketing, sales, customer support, admin - include only what applies). 5-10 questions per section, 20-40 questions total.

Questions should uncover: repetitive/manual work, bottlenecks, data usage gaps, automation opportunities.

Be concise and practical. Avoid buzzwords. Focus on uncovering real problems, not theoretical ideas.

AI implementation roadmap

Dump your audit findings into this prompt. It generates a prioritised plan you can present back to the client or use internally.

You are a senior AI implementation strategist. Translate these business audit findings into a practical, ROI-driven AI solutions roadmap.

Business type: [INSERT BUSINESS TYPE]
Key pain points: [INSERT PAIN POINTS]
Audit notes: [PASTE AUDIT RESPONSES]

Output structure:
1. Key Insights - summarise critical inefficiencies and where time/cost is being lost
2. Prioritised AI Opportunities - ranked list. For each:
   - Opportunity name
   - What it solves
   - Proposed AI solution
   - Estimated time savings
   - Implementation difficulty (Low/Medium/High)
   - Speed to implement
   - Expected business impact
3. Quick Wins - top 3-5, lowest effort + immediate ROI
4. Execution Roadmap:
   - Phase 1: Quick wins (0-30 days)
   - Phase 2: Mid-level (1-3 months)
   - Phase 3: Strategic/advanced (3+ months)

Be practical. Tie everything to the audit inputs. Assume limited technical resources.

Web scraping for lead lists

Setup: the Apify connector in Claude Cowork. Connect via the Connectors tab and paste your Apify API key (free account at apify.com).

Local business scrape

Use Apify to scrape 20 [INDUSTRY] companies in [LOCATION]. For each, collect: business name, website, contact person (if available), email address, phone number, and location. Put everything into a clean spreadsheet.

Enrich with decision makers

Take the list of [INDUSTRY] companies we just scraped. For each one, use vibe prospecting to find the key decision makers - founders, directors, or managing directors. Get their names, job titles, LinkedIn profiles if available, and any useful contact information. Add this to the spreadsheet alongside the company data.

Job description scrape for research

Use Apify to scrape LinkedIn for [NUMBER] [ROLE TYPE] job listings in [LOCATION]. Pull the job title and full job description for each listing.

Research report from scraped data

Analyse all the job descriptions we scraped. I want to know:
1. How many mention AI skills as a requirement (ChatGPT, Claude, prompt engineering, etc.)
2. Break it down by industry/sector
3. Break it down by seniority level
4. What specific AI skills are mentioned most often
5. Any patterns or trends worth highlighting

Turn this into a short research report that could be used for a press release. Include key stats, summary of findings, and 2-3 quotable takeaways.

Cowork use cases: 5 prompt templates

Hiring manager

I'm the hiring manager at [Company Name]. I need you to work through our [Role Title] recruitment folder, take the raw applications, build a scoring framework based on the job description, score each candidate, and create a leadership summary spreadsheet with our top candidates highlighted and ready for interview decisions.

SEO auditor

Can you look at the Search Console data in the [folder name] folder and understand the performance of my website? Then navigate to the top 5 performing pages using the web browser, look through them, and find areas that can be improved - outdated information, content gaps, SEO issues. Report back with a list of actions I need to take on each page.

Customer intelligence builder

Can you go into my Gmail - there's a folder called [folder name]. Read all those emails and bring all the information into a single report. Then using that information, build me a customer ICP report as a Word document and an interactive HTML intelligence dashboard. I want it to show key insights into my customers as a pattern - their pains, their dreams, their goals, and what content I should be creating.

Content strategist

Can you get all of the links for my YouTube competitors from this Notion page? Then browse and research their channels - look for trending topics, top-performing recent content, and patterns I could take advantage of. Cross-reference this with the ICP research we've already done. Then create a Notion content pipeline database with a 30-day content plan based on 3 YouTube videos a week. Each week should have a specific topic or theme.

Group travel agent

I've got a group [trip type] to [destination]. There's [number] of us going. I've given you access to all the documentation for the different bookings - flights, hotels, [activity] rental, restaurant reservations, etc. Go through it all and create a day-by-day itinerary I can share with the group so I don't have to keep answering questions in the group chat. Include everyone's travel details, timings, and any payments still owed.

Build a Claude skill from any prompt

If you've typed the same prompt more than twice, turn it into a skill. Enable the Skill Creator in Settings → Customise → Skills first, then use /skill in any Cowork session.

I want to build a Claude skill. I have a [NAME] prompt I use regularly. I want to turn it into a Claude skill so that any time I need [TASK], I can just say '[TRIGGER PHRASE]' and it automatically runs the process without me having to re-explain anything.

Here's my existing prompt:
[PASTE YOUR PROMPT]

Take this prompt and enhance it into a repeatable skill that's clear, structured, and easy to activate.

Decision council skill

Spawns 5 expert personas in parallel (Researcher, Sceptic, Strategist, Operator, Creative). They evaluate your question independently, run an anonymous peer review, and a chairman synthesises everything into a recommendation. Output is an interactive HTML dashboard.

I want to build a decision-making council skill. It should be a multi-persona decision-maker that can help me with any questions and decisions. The skill should spawn five parallel expert personas: a Researcher (evidence-first), a Sceptic (stress-tests the decision), a Strategist (long-term view), an Operator (execution focus), and a Creative (lateral thinker).

Process: each agent evaluates the question independently. Then an anonymous peer review round where each agent responds to the others. Then a chairman-style synthesis that acknowledges key tensions, identifies the strongest 2-3 options, makes a recommended next step, and flags deal-breakers.

Output: interactive HTML report with 5 council member cards (name, role, verdict, 3 bullet points), peer review section, and chairman synthesis box with recommendation highlighted.

Build this as a Claude skill with trigger phrase "council this".

Framework reverse engineer skill

Takes any expert content (video, podcast, article, book chapter) and extracts the hidden methodology, names the principles, and builds it into a reusable framework.

I want to build a Claude skill. I have a link to expert content about [TOPIC]. I want to extract [EXPERT NAME]'s expertise and use that to build a skill I can run any time I want to reverse-engineer and build a framework from expert content.

Here is a transcript of the content:
[PASTE TRANSCRIPT]

Build this into a skill that: reads the content, extracts the hidden methodology, names the principles as a proprietary framework, defines when to use it, and shows the step-by-step application with real examples.

Brand applicator skill

Applies your exact brand colours, fonts, logos, and tone to any output — carousels, Word docs, decks, reports, invoices, dashboards — with a single phrase.

I want to build a brand applicator Claude skill. Act as a brand designer. I've attached my brand files - logos, colours, fonts, and brand guidelines.

Extract everything you need: logos, icon variations, exact colour hex codes, font specifications, tone of voice rules, layout preferences, and any do/don't guidelines.

Then build a Claude skill called Brand Applicator that I can activate by saying "make this branded". The skill should be able to apply my brand to any output: LinkedIn carousels, Word documents, PowerPoint presentations, PDF reports, invoices, React dashboards - anything I produce.

Include all extracted brand information directly in the skill so Claude always has it available when the skill is active.

Upload your logo, brand colours, and any brand guidelines doc alongside this prompt.

ChatGPT to Claude migration

Export your ChatGPT data (Settings → Data controls → Export data), download the chat.html file, upload it to a Claude Project or Cowork session, then run this prompt.

You are a senior AI cognitive migration architect. Analyse my complete ChatGPT conversation history and reconstruct it into a structured knowledge base.

Produce sections covering:
1. Identity Profile - personality traits, cognitive style, communication preferences, motivations
2. Long-Term Vision and Goals - explicit and implicit goals, recurring ambitions
3. Active and Recurring Projects - name, purpose, current state, open loops, key decisions, constraints
4. Knowledge Domains and Expertise Map - areas of competence, tools frequently discussed, technical stack
5. Decision-Making Patterns - risk tolerance, optimization style, iterative vs perfectionist tendencies
6. Behavioral Patterns - repeated questions, shifts in interest, signs of scaling intent
7. Constraints and Operating Conditions - time, resources, skill gaps
8. Preferred Output Styles - formatting, tone, level of detail
9. Open Loops and Unresolved Threads - ideas mentioned but not developed
10. Memory Compression Layer - highly compressed, maximum signal density, only durable traits and active vectors, designed for long-term memory injection

RULES: do NOT summarise chronologically. Do NOT restate entire conversations. Do NOT include filler. Distill repetition into patterns. If something is ambiguous, label it as inferred.

OUTPUT: Clean markdown headings and bullet points. No JSON. No meta commentary. Only the structured knowledge base.

Build a professional website in 20 minutes: 4-prompt chain

Run all 4 prompts in one conversation thread. The final output is a ready-to-paste Lovable.dev prompt. Take screenshots of 1–2 websites you like before starting.

Prompt 1: business context

I'm building a website and need help getting clear on what I need.

What we do: [Your answer]
Who it's for: [Your answer]
Main visitor action: [Book a call / buy / download / contact]
What makes us different: [Your answer]
Why this business exists: [Your answer]

Create a clear website brief including: core messaging (homepage headline + subheadline), essential pages, key homepage sections, and suggested CTA buttons.

Prompt 2: extract design inspiration

[Upload screenshots of websites you like]

Analyse these images and extract: layout structure, visual style (typography, spacing, image usage), content patterns, interactive elements, colour palette, and overall design principles that make it work.

Present as a structured breakdown I can reference when building. Don't copy the content - extract the design DNA.

Prompt 3: brand identity

Now add my brand identity:

Brand Colours:
- Primary: [hex + name]
- Secondary: [hex + name]
- Accent: [hex + name]
- Background: [hex + name]
- Text: [hex + name]

Typography:
- Headings: [font name]
- Body text: [font name]

Brand Voice: [how you communicate - e.g. "Direct and conversational"]

Take the website structure from earlier and adapt it with MY brand elements. Show how my colours, fonts, and voice would transform that design inspiration into something uniquely mine.

Prompt 4: generate the Lovable spec

You now have my business context, design inspiration, and brand elements.

Build me a comprehensive Lovable.dev prompt that combines all three. Specify: pages and structure, exact sections for each page, specific colour hex codes and where to use them, typography choices, layout patterns, interactive elements and CTAs, responsive design considerations, navigation structure.

Format as a single copy-paste prompt I can drop into Lovable to generate my website in one go.

Take the output to Lovable → New Project → paste the prompt.

Run all 4 prompts in one conversation, in order. Set aside 60–90 minutes. The more honest your answers in Prompt 1, the more useful the output from Prompts 2–4.

Prompt 1: reflect

You are a thoughtful personal development coach guiding me through my annual review. Work through this in batches. Wait for my responses before continuing.

Batch 1: Review
1. What were my 3 biggest wins this year?
2. What were my 3 biggest lessons (including failures)?
3. What new habits did I build? Which stuck?
4. Where did most of my time actually go?
5. What made me happiest this year?
6. What drained my energy the most?
7. Which relationships grew stronger? Which faded?
8. What goals did I set at the start of the year? How did they turn out?
9. What am I most grateful for?
10. What do I wish I had done differently?
11. What did I spend money on that was worth it? What wasn't?
12. Where was I out of alignment with my values?

Batch 2: Memories
Complete these sentences:
1. The best surprise of the year was...
2. The best experience I had was...
3. The strongest emotion I felt was...
4. The best day I had was...
5. The quote that stuck with me was...

Batch 3: Plan
1. When I'm 85, what will I wish I'd spent more time on?
2. If I could only improve ONE thing about my daily routine, what would it be?
3. What's the biggest time-waster I need to eliminate?
4. Describe my ideal ordinary Tuesday next year.
5. What conversations am I avoiding that I need to have?
6. What would I attempt if I knew I couldn't fail?
7. What skill would most accelerate my goals?

Final questions:
1. How will I know next year was a success? (Be specific)
2. Who do I need to become to achieve what I want?
3. What advice would I give myself at the start of this year?

Prompt 2: extract patterns and blind spots

You are an expert pattern recognition analyst. Analyse my annual review responses and surface insights I may not see myself.

1. Key Lessons - 5-7 most important lessons and why they matter for my future
2. Patterns - recurring behaviours (positive and negative), themes in what energised vs drained me
3. Insights - what story is my year telling me about what I truly value, where I'm aligned vs misaligned
4. Blind Spots - be direct and honest. Contradictions in my answers, things I'm avoiding, patterns I might be rationalising, uncomfortable truths. Don't be gentle here. I need the aha moments.
5. Summary Statement - 2-3 sentences on the core message my year is trying to tell me

Reference my actual answers, not generic observations.

Prompt 3: envision

Based on my reflections and your analysis, help me build a clear vision for next year.

1. Who do I need to become? Not what I need to do - who I need to BE.
2. What does success look like? If I'm sitting here 12 months from now having crushed it - what's different? What does my typical day look like, what have I achieved, how do I feel?
3. What must I protect? What's already working that I should guard?
4. What must I release? What beliefs, habits, or patterns need to go?

Keep it specific to me - reference my actual responses and the patterns you identified.

Prompt 4: roadmap and dashboard

Turn everything we've discussed into an actionable roadmap.

1. My Goals (3-5) - aligned with who I need to become and the vision we built. For each: the goal (specific and measurable), why it matters, what success looks like by end of year.

2. Quarterly Milestones - for each goal, what should be true at Q1/Q2/Q3/Q4?

3. Keystone Habits - 3-5 daily or weekly habits with highest impact. For each: the habit, when/how to do it, which goals it supports.

4. Stop Doing List - 5-7 things to actively stop or avoid based on patterns and blind spots.

5. One-Page Annual Plan:
THEME: [one word or short phrase]
WHO I'M BECOMING: [2-3 sentences]
MY CORE GOALS: [goals + success metrics]
KEYSTONE HABITS: [daily / weekly]
NOT DOING: [top 3]
Q1 FOCUS: [primary focus for first 90 days]
SUCCESS CRITERIA: [how I'll know the year was a win]

6. Interactive Goal Dashboard - create a React dashboard I can use as a daily tracker. Include: year theme, today's date and days remaining, year progress bar, goals progress (current vs target with progress bars and trend indicators), quarterly milestone checkpoints, habits tracker with weekly grid and streak counter, current quarter focus panel, stop-doing list as reminders. Dark mode. Scannable in under 5 seconds.