A hands-on workshop to build AI-powered GTM systems
Visit 100x.productgrowth.blog to access this presentation
Analyze 50+ competitor LinkedIn ads in 30 minutes. ChatGPT's Agent mode browses LinkedIn Ad Library, captures creatives, and delivers strategic analysis on messaging patterns, hooks, and positioning.
"I just did in 30 minutes what my team quotes 2 weeks for."
Go to linkedin.com/ad-library and note down 3-5 competitor company names you want to analyze. Make sure they're actively running LinkedIn ads.
Use ChatGPT Pro or Plus with the "Browse with Bing" or "Computer use" capability enabled. This allows the AI to navigate web pages autonomously.
Use this prompt: "Go to LinkedIn Ad Library. Search for [COMPETITOR NAME]. Analyze their top 10 ads. For each ad, note: headline, hook, CTA, visual style, and value proposition. Then provide strategic insights on patterns you observe."
Run the same analysis for each competitor. ChatGPT will browse, screenshot, and analyze each ad. Export the results to a comparison document.
Ask ChatGPT: "Based on all competitor ads analyzed, what are the 5 key messaging gaps I can exploit? What hooks are overused? What positioning is nobody taking?"
Use Perplexity first to identify which competitors are running the most ads, then prioritize those for deeper ChatGPT analysis.
Build a prompt system that lets you have instant "conversations" with simulated buyers. Test messaging, understand objections, and refine positioning—all without scheduling a single customer call.
"I just had a conversation with my buyer without scheduling a call."
Write down: Job title, company size, industry, main responsibilities, KPIs they're measured on, tools they use daily, and their reporting structure.
Use this template: "You are [TITLE] at a [SIZE] [INDUSTRY] company. Your main challenges are [X, Y, Z]. Your boss cares about [METRIC]. You're skeptical of vendors. Respond to everything as this persona would—be realistic, push back on vague claims."
Share your current pitch, landing page copy, or email sequence. Ask the persona: "What's your honest reaction to this? What would make you ignore this? What would make you take a meeting?"
Ask: "What are your top 5 objections to a product like ours? For each objection, what evidence or proof would overcome it?"
Save the persona prompt as a template. Run different messaging variations through it. Build a "persona testing library" for your entire ICP matrix.
Add real customer call transcripts to your prompt to make the persona more authentic: "Here are 3 real conversations with this persona—learn from their language, concerns, and decision-making patterns."
Turn call transcripts, sales conversations, and customer feedback into high-signal content ideas. Mine your existing conversations for content gold that actually resonates with your audience.
"I've been sitting on 50 call recordings and never thought to mine them for content."
Export 3-5 call transcripts from Gong, Fireflies, or any recording tool. Focus on: discovery calls, demo calls, and churn conversations—these contain the richest insights.
Upload transcripts to ChatGPT. Use this prompt: "I'm uploading customer call transcripts. Your job is to extract content ideas that would resonate with prospects like these. Focus on: pain points mentioned, objections raised, 'aha' moments, and questions asked."
Ask: "From these transcripts, give me 10 content ideas with these formats: 3 LinkedIn posts, 3 blog post titles, 2 email subject lines, 2 webinar topics. For each, include the specific customer quote that inspired it."
Ask: "What are the top 3 themes or concerns that appear across multiple calls? How might we create a content series around each theme?"
Create a simple spreadsheet with: Idea, Format, Source Quote, Priority, and Status. This becomes your content backlog fed directly from customer voice.
Add this to your prompt: "Flag any surprising insights or counterintuitive statements—these make the best 'contrarian' content pieces that get high engagement."
Build a Custom GPT that takes calendar context, email threads, and LinkedIn profiles to produce a complete meeting prep brief. Get all context in 2 minutes instead of 15 minutes of scrambling.
"I'll use this before every external meeting starting tomorrow."
Go to chat.openai.com/gpts/editor. Name it "Meeting Prep Assistant" and add a description: "Prepares comprehensive briefs for upcoming meetings."
"You are a meeting preparation expert. When given information about an upcoming meeting, you produce a structured brief including: Person background, Company context, Recent news, Likely agenda, Questions to ask, Potential objections, and Talking points. Be concise and actionable."
In the GPT settings, enable "Web Browsing" capability. This allows the GPT to look up company news, LinkedIn profiles, and recent developments in real-time.
Upload your company's pitch deck, battlecards, and common objection handlers as knowledge files. This helps the GPT tailor talking points to your specific offering.
Input: "I have a meeting with [NAME], [TITLE] at [COMPANY] tomorrow. Here's the email thread: [PASTE]. Prepare my brief." Iterate on the output format until it matches your workflow.
Add a section in your instructions for "Icebreakers"—the GPT can find shared connections, mutual interests, or recent personal updates from LinkedIn to help start conversations warmly.
Build a Custom GPT trained on proven cold email frameworks. Upload your ICP, messaging, and brand voice—walk out with a personalized outbound writing assistant your entire SDR team can use.
"My SDRs are going to love me for this."
Gather 10-15 of your highest-performing cold emails. Include reply rates and context. These become the "golden examples" your GPT will learn from.
Write out your framework: "1. Pain-focused opener (reference specific challenge), 2. Bridge (how we solve it), 3. Social proof (specific result), 4. CTA (low-friction ask)." Add anti-patterns: what to avoid.
System prompt: "You are an expert cold email copywriter. You write emails that are: under 100 words, focused on the prospect's pain, specific not generic, and have clear low-friction CTAs. Never start with 'I', never use 'reaching out', never be salesy."
Add as knowledge files: ICP document, persona pain points, case studies with specific metrics, competitor comparison points, and your list of banned phrases/patterns.
Define what users should input: "To generate an email, provide: Prospect name, Title, Company, Industry, One specific insight about them (from LinkedIn/news)." This ensures consistent quality.
Add a "critique mode" to your GPT: "After writing each email, critique it against these criteria and suggest one improvement." This builds in quality control.
Upload 20-30 call transcripts into NotebookLM and get strategic pattern analysis AND an AI-generated audio briefing you can share with leadership. Saves 2-3 hours versus 15+ hours of manual analysis.
"I just created an executive briefing from raw calls in an hour."
From Gong, Fireflies, or your recording tool, export 20-30 call transcripts. Focus on a specific call type: discovery, demo, or closed-won/lost for cleaner insights.
Go to notebooklm.google.com. Create a new notebook named "Sales Call Analysis - [Month/Quarter]". Upload all transcripts as sources.
Ask NotebookLM: "What are the top 5 objections across all calls? What pain points are mentioned most? What competitors come up and why? What questions do prospects ask most?"
Click "Audio Overview" in NotebookLM. It generates a podcast-style briefing that summarizes all your sources. This is the "holy shit" moment—share it with your CEO.
Final query: "Based on all these calls, what are 5 specific recommendations for our sales team? What should we change in our pitch? What objection handling needs improvement?"
Create separate notebooks for won vs. lost deals. Compare the two: "What did we say in won calls that we didn't say in lost calls?" This reveals your actual differentiators.
Build a Claude project that simulates customer reactions to your marketing before launch. Feed it real customer data, then A/B test campaigns against it without waiting for real-world data.
"I can now A/B test messaging without waiting for real data."
Go to claude.ai and create a new Project called "Customer Digital Twin". This gives you persistent context across conversations.
Add to Project Knowledge: 10-20 customer call transcripts, NPS feedback, support tickets, and G2/Capterra reviews. This trains Claude on how your real customers think and speak.
"You are a simulation of our typical customer. Based on all the uploaded customer data, you think, react, and respond exactly like our real customers do. When I show you marketing materials, respond as a skeptical but potentially interested customer would."
Share a landing page, email, or ad copy. Ask: "What's your gut reaction? Would you click? What questions do you have? What would make this more compelling?"
Share two versions: "Here are two subject lines. As our customer, which would you open? Why? Which feels more relevant to your actual problems?" Document responses and iterate.
Create multiple digital twins for different segments. A "startup founder" twin will react differently than an "enterprise VP" twin. Test messaging against each to refine targeting.
Build a Custom GPT loaded with your actual company docs—positioning, personas, battlecards, tone guidelines. Walk out with the "smartest new GTM hire" that needs no onboarding.
"I just built my replacement... and I'm okay with it."
Collect: Messaging framework, ICP document, buyer personas, competitive battlecards, brand voice guidelines, product one-pager, case studies, FAQ document. Convert all to PDF or text.
Name: "[Company Name] GTM Operator". Description: "Your company's GTM expert. Knows our positioning, personas, competitors, and can create on-brand content."
"You are the GTM expert for [COMPANY]. You know our product, positioning, and customers deeply. When asked to create content, always align with our messaging framework. When asked about competitors, reference our battlecards. Never make up features—only reference what's documented."
Upload every document as knowledge. Enable "Code Interpreter" for data analysis tasks. Enable "Web Browsing" to research competitors and stay current.
Add conversation starters: "Write a cold email to a [PERSONA]", "Create LinkedIn posts about [FEATURE]", "How do we position against [COMPETITOR]?", "Draft a one-pager for [USE CASE]"
Share the GPT with your entire team. As you get feedback, update the knowledge files. The GPT becomes your team's shared brain that gets smarter over time.
Using bolt.new, turn a boring Excel ROI model into a beautiful, interactive web app your sales team can use with prospects. Visual transformation is dramatic—what agencies quote $10k for.
"My HubSpot agency quoted $10k for this. I just built it in an hour."
Write out your Excel formulas in plain English: "Time saved per week × hourly rate × 52 weeks = annual savings. Implementation cost ÷ annual savings = payback period." Include all variables and assumptions.
Navigate to bolt.new. This is an AI-powered web app builder. You'll describe what you want, and it generates working React code instantly.
"Build an interactive ROI calculator for [PRODUCT]. Inputs: [list your variables with ranges]. Calculations: [paste your formulas]. Outputs: Monthly savings, Annual savings, Payback period, 3-year ROI. Make it look professional with a modern dark theme and animated number transitions."
Ask: "Add a comparison slider between 'without us' and 'with us'. Add a chart showing savings over 36 months. Add a 'Download as PDF' button for proposals."
Click "Deploy" in bolt.new to get a live URL. Share with your sales team. Optionally, embed it on your website's pricing or ROI page.
Add lead capture: "When someone enters values, show a modal asking for their email to save their calculation and receive a detailed PDF report." Now it's a lead gen tool.
Build a Claude Project with layered context—past content, brand voice "skills," content pillars, audience insights. This is the system behind 1M+ monthly LinkedIn impressions for top creators.
"This isn't a prompt. This is an entire content operation."
In claude.ai, create a new Project called "Content Strategist". This becomes your persistent content creation workspace with memory across sessions.
Add to Project Knowledge: Your top 20 performing posts/articles, brand voice guide, content pillars document, audience research, and examples of content styles you admire.
"You are my content strategist. You know my voice, my audience, and my content pillars intimately. When I give you an idea, you help develop it into a compelling piece. You push back on weak angles. You suggest hooks. You edit ruthlessly. My pillars are: [LIST]. My audience is: [DESCRIBE]. My voice is: [CHARACTERISTICS]."
Build reusable prompts: "LinkedIn post from this idea", "Turn this into a thread", "Write the counterintuitive take on this", "Create 5 hooks for this topic", "Edit this draft to be more punchy".
After each piece performs, add notes to the Project: "This hook worked because..." or "This format underperformed because...". The AI learns your audience's preferences over time.
Create a "swipe file" document of 50+ great hooks from creators you admire. Upload it as knowledge. Ask Claude: "Write 5 hooks for this topic in the style of my swipe file examples."
Pick 1-2 use cases and create something you'll actually use Monday
Want more use cases?
Check out the full list →Questions? Flag us down. We're here to help you build.