Get Unlimited Free Leads With Claude

Turn Claude into your own expert list builder that scrapes any website, finds the contact details, and builds a full lead list for you, all run from Claude Code in plain English. No new subscriptions, no code, no agency. Layer 1 is the setup. Layer 2 is the verify step that makes scraped leads actually convert.

What You'll Learn

• The free tool that replaces Apify, Apollo, and lead-gen VAs: what ScrapeGraphAI is and why MIT-licensed and free forever matters for you
• The 3-minute setup: the exact install commands and how to run the whole thing from Claude Code in plain English
• The plain-English scraping pattern: how to describe what you want so the AI returns clean contact data instead of garbage
• Layer 2: the verify-enrich-segment pass that takes a raw list from a 12% bounce rate to 90%+ deliverable
• The four highest-leverage use cases (lead lists, competitor products, real estate listings, job listings) plus the failure modes you will hit

What This Guide Is (And Isn't)

This is the setup and the playbook behind turning Claude into a free lead-scraping machine, not a one-click magic button. By the end you will have it running from Claude Code, and you will understand the part nobody talks about: how to turn raw scraped data into leads that actually convert.

Why two layers. Most scraping guides stop at "here is how to pull data." That is Layer 1. The reason 90% of scraped leads are garbage is that people skip Layer 2, the verification and enrichment that makes the data usable. This guide covers both.

Note: this is a setup and playbook overview, not legal advice. Scrape responsibly and respect each site's terms.

The Setup That Replaces Apify, Apollo, And VAs

ScrapeGraphAI is a free, open-source tool that puts an AI model in charge of the scrape. Instead of writing brittle selectors or paying for a platform, you point it at a site and describe what you want in plain English. Wire it into Claude Code and Claude becomes an expert list builder: it scrapes any website, finds the contact details, and hands you a finished lead list.

Why it replaces the paid stack. Apify charges per usage and per actor. Apollo charges a seat fee for a database that goes stale. A lead-gen VA charges hourly to copy-paste what this does in seconds. ScrapeGraphAI is MIT licensed, which means free forever and commercial use allowed, and you can run it fully local for zero cost per lead. Over 26,000 developers have starred it.

What it is not. It is not a hosted button with no setup. You run it on your own machine through Claude Code. That is the trade: you skip the monthly subscription, you do a one-time setup. This guide is that setup.

Layer 1: The 3-Minute Setup In Claude Code

You need Python 3.10 or newer and one API key for the model that does the reading (or run it fully free with local Ollama). The fastest path is to let Claude Code do the install for you: open Claude Code and tell it "install ScrapeGraphAI and walk me through it." If you would rather run the commands yourself, here they are:

# 1. Install the library
pip install scrapegraphai
playwright install

# 2. Set your model API key (or skip this and use local Ollama for free)
export OPENAI_API_KEY="your-key-here"

# 3. Confirm it imported
python -c "import scrapegraphai; print('ready')"

The vibe-coding path. You do not need to be technical. Paste your goal into Claude Code in plain English ("scrape every personal injury firm in Ontario with their partners and phone numbers") and Claude writes and runs the scrape for you. You describe the outcome, Claude does the typing.

The Plain-English Scraping Pattern

This is the whole point. You are not writing scraping code, you are writing a sentence. The structure is always the same: name the URL, name the fields you want, and be specific about the output. Here is the minimal pattern Claude runs under the hood:

from scrapegraphai.graphs import SmartScraperGraph

scraper = SmartScraperGraph(
    prompt="List every business on this page. For each return: name, website, phone,
    and email. Skip any row without an email.",
    source="https://example.com/directory",
    config={"llm": {"model": "openai/gpt-4o-mini"}}
)
print(scraper.run())

The specificity of your prompt is what separates clean data from garbage. "Get the leads" returns mush. "Return name, company, role, and email, skip any row without an email" returns a usable list.

The specificity rule. Always name the exact fields and always name the exclusion. The single biggest quality jump comes from adding "skip anything missing X" to the end of your prompt.

Layer 2: Make The Data Actually Convert

This is the layer that separates people who get results from people who burn their domain. Scraping is step one. A raw scraped list is full of duplicates, generic info@ inboxes, and dead addresses. Send to it cold and you torch your sender reputation. The fix is a three-step routing pass.

The verify, enrich, segment pass:

1. Verify: run every email through a validation check and drop anything that does not resolve.
2. Enrich: fill in the missing role, company size, or recent activity so you can personalize.
3. Segment: split the list by the one variable that changes your message: industry, size, or intent signal.

The number that matters. Industry data puts scraped-list bounce rates around 12% before cleaning. After a verification pass you want 90%+ deliverable. Emails to a named decision-maker convert 5 to 10x better than generic inboxes. The cleaning is the work.

What To Actually Pull (4 High-Leverage Use Cases)

The tool is generic. The leverage is in what you point it at. The four that earn their keep:

1. Lead lists. Point Claude at a directory or map results and pull every business or decision-maker in a niche with website, phone, and contact. This is the $5,000 agency deliverable, run in an afternoon.
2. Competitor product and price tracking. Pull every competitor's product, price, and review into one spreadsheet on a schedule. A weekly manual slog becomes one sentence.
3. Real estate listings. Scrape listings for price, beds, location, and agent contact, the same data investors pay subscriptions to monitor.
4. Job listings. Pull open roles at the companies you want to work for, with the hiring contact, so you can reach the decision-maker directly instead of the application black hole.

Stay clean. Respect each site's terms of service and robots rules, avoid scraping behind logins or personal data you have no right to, and throttle your requests. The goal is leverage, not a cease-and-desist.

Failure Modes You Will Hit

Forewarned is forearmed. The four that trip everyone up, and the fix for each:

1. The site blocks you. Heavy anti-bot sites need slower request rates or a proxy. Start with public, scrape-friendly sources before you fight CloudFlare.
2. The output is mush. Your prompt was vague. Name the exact fields and add the exclusion rule from the pattern above.
3. The list is full of garbage. You skipped Layer 2. Run the verify, enrich, segment pass before you send anything.
4. Names hide behind JavaScript. Some team pages load names dynamically. Point Claude at the individual profile pages, or map the site's profile URLs and pull names from those.

Build It Yourself This Weekend

If you can follow the 3-minute setup, you have everything you need to stand up a working scraper today. Run it from Claude Code so you can trigger the whole thing in plain English, and you have a personal lead-scraping agent that costs you nothing per month. Your checklist:

• Python 3.10+ installed and a model key set (or local Ollama for free)
• ScrapeGraphAI installed, the "ready" check passes
• One successful scrape from Claude Code with named fields and an exclusion rule
• A verify, enrich, segment pass on the output
• The whole flow triggered from one plain-English Claude Code prompt

Go Deeper

Printable PDF version of this guide: https://drive.google.com/file/d/18pnyyE9s9lvAu8vzBe17hgfmOJ0KPTjc/view?usp=drivesdk

Want the whole system installed live, with Cooper on the call? Build Your Own AI, the two week sprint runs the first two weeks of August.

Or get a new working tool every Monday inside The AI Operator community on Whop.