Hi, I'm Nicolas

nico.webp

I spent the last 15 years writing code, and back in 2016 I built OceanWP, a WordPress theme that ended up running on more than 500,000 websites. I sold it in 2019, took a long break to figure out what I actually wanted to build next, and I'm now back with my wife Maria building our next SaaS together.


Now, the 4-prompt Claude playbook

This playbook is a sequence of 4 Claude prompts. You paste the inputs, Claude produces real working output, you take the output to the next step. The whole thing takes about 30 minutes from a blank ICP brief to 25 personalized DMs ready to send on LinkedIn.

Industry data shows personalized signal-based LinkedIn outreach hits 25-40% reply rates compared to 1-2% on cold lists. The system below is built to consistently produce the kind of message that lands inside that 25-40% band.

Tools you need: a LinkedIn account (free tier works, Sales Navigator is faster), Claude Opus 4.7 access (Claude.ai, Anthropic API, or any tool that lets you bring your own key), and 30 minutes.


STEP 1: Build your ICP brief (5 minutes)

The whole system is anchored on a precise ICP description. Vague ICP equals vague output equals 1% reply rates. Spend the 5 minutes here and the next 25 minutes deliver.

<aside> ⚠️ This step outputs a STRUCTURED BRIEF, not Boolean queries. The Boolean queries get generated in Step 2. If Claude returns Boolean queries here, the prompt has been mis-followed and Step 2 will produce broken URLs.

</aside>

Open Claude and paste this prompt. Replace the bracketed parts with your actual specifics:

You are an ICP definition expert. Help me write a clean structured ICP brief that I will paste into a separate prompt for LinkedIn search URL generation.

CRITICAL: Output ONLY the structured brief in the format below. Do NOT output Boolean queries. Do NOT combine fields with AND/OR. Do NOT generate URLs. Just the raw structured brief, one item per line where listed.

I sell: [describe your product or service in one sentence]

My best 2 to 3 past or ideal customers: [describe them by role, company stage, what they were trying to do when they bought]

The main pain my product solves: [the specific pain in 1 sentence]

Output format (use exactly these 5 sections, one item per line in lists, no Boolean syntax anywhere):

1. JOB TITLES (5 to 7 titles, one per line, no quotes, no parentheses, no AND/OR)
- Title 1
- Title 2
- ...

2. INDUSTRIES (3 to 5 industries, one per line, broadest first)
- Industry 1
- Industry 2
- ...

3. COMPANY SIZE (one line, employee count range)
- e.g. 11 to 200 employees

4. COMPANY STAGE (one line, pick relevant)
- e.g. early stage to growth (Seed to Series B)

5. HEADLINE SIGNALS (3 to 5 phrases the prospect uses in their LinkedIn headline when they are in-market, one per line, no quotes, no AND/OR)
- e.g. scaling team
- e.g. building from zero
- e.g. raised seed

End of brief. Stop here. Do not append Boolean queries, examples, or LinkedIn search strings. Those come from the next prompt.

Output you will get: a 5-section brief with plain text items per line. You save it somewhere accessible. You paste this brief into Step 2 prompt as the ICP context.

<aside> ✅ Concrete example output Claude should return:

  1. JOB TITLES
  1. INDUSTRIES
  1. COMPANY SIZE
  1. COMPANY STAGE
  1. HEADLINE SIGNALS

</aside>


STEP 2: Generate 5 ready-to-click LinkedIn search URLs (1 minute)

<aside> 👉 How this step works in plain English: you paste the prompt below into Claude. Claude returns 5 full LinkedIn URLs. You copy each URL and paste it into your browser address bar. Each URL opens directly to a filtered LinkedIn People page. That is it. No encoding, no clicking through filters, no search bar.

</aside>

Paste this prompt into Claude: