Source: raw/How_to_Generate_Unlimited_leads_with_1_Claude_Skill_free_template.md (primary); raw/8_Claude_Skills_to_Automate_Your_Sales.md (lead-gen, LinkedIn post engager and CRM prospect mining skills)
Creator: Ben AI · URL: https://www.youtube.com/watch?v=kXAp1YzPcVI · Duration: 15:58 · Platform: YouTube · Uploaded: 2026-10-08
Second video: https://www.youtube.com/watch?v=pr5Pce7Z4eg · 16:37 · uploaded 2026-08-03
Ben AI runs an AI agency and sells a paid “AI accelerator” community; both videos give the skills away free through a Notion page and push the paid tier. The useful part is his four-decision framework for building one outbound skill per campaign: where the leads come from, how they are qualified, what enrichment they need, and a human-written message template that Claude fills in. Each campaign becomes its own skill, so lead sources and messages can be tested against each other. Auto-caption fixes applied: “Appify/Ampify” → Apify, “VIP/Vipe prospecting” → Vibe Prospecting, “any mailinder” → Anymail Finder ^[inferred], “Lamb list” → lemlist ^[inferred], “clot/cloud” → Claude.
Key Takeaways
- A “skill-creator for lead-gen skills”. Ben’s free skill does not scrape leads directly. It interviews you and then builds a customized lead-gen skill for one niche and one ICP. His framing is that there is no one-size-fits-all outbound skill, because the lead source and the message depend on the niche.
- One skill per ICP per lead source. If you have several customer profiles, build one skill each and treat each as an experiment. Ben built three: fintech-unicorn sales leaders, SaaS companies hiring for sales roles, and people who engaged with his LinkedIn posts. He says the hiring-signal skill is “performing really well, a lot better than the other one”, so he is putting more into it. No reply rates are given.
- The lead source matters most. Where you scrape decides which connector you need (table below). Ben’s default advice: install Apify and Firecrawl whatever your ICP, because “you can scrape basically 90% of the internet through these two.”
- Qualify with an extra research step. Lead databases can pre-filter by size, geography or revenue, but not on criteria like “sales team over 25 people”. The skill adds a research pass after scraping to check those. Ben prefers “highly targeted smaller lead lists” to “spray and pray”.
- Enrichment has three parts: the right job titles to contact (his: VP of Sales, Head of Sales, Sales Director), contact details for your channel (an email finder for email outreach; he uses Anymail Finder, which he calls cheap and verified), and research data points for personalization.
- Do not let Claude write the outreach from scratch. Ben: “it’s just not that good at it.” Write the template yourself with personalization variables, and let Claude fill the variables from its research.
- Output is a sheet, not a send. Each run produces a Google Sheet and CSV (lead, contact, qualification data, personalized message) that he uploads to Instantly to send. Sending stays outside Claude.
The four decisions before you build
1. Lead source → connector
| Customer type | Where they are | Connector Ben uses |
|---|---|---|
| Local businesses | Google Maps; sometimes Google search, niche listings, directories | Apify (Google Maps scrapers) |
| Tech, SaaS, B2B | LinkedIn, Sales Navigator, lead databases (Apollo, lemlist) | Apify (LinkedIn and Sales Navigator scrapers), Apollo |
| Signal-based B2B | Companies showing hiring, funding, tech-stack or leadership-change signals | Vibe Prospecting |
| Influencers and creators | Social platforms | Apify |
| Small niches | Directories, aggregators, Google results | Firecrawl |
Vibe Prospecting is described as an AI-first lead database exposed as an MCP connector, with intent signals an agent can query. Ben recommends it for non-local B2B targets and says its cost “scales with how much you use it” and that it almost always returns the email already.
2. Qualification criteria
List which criteria the source can filter on and which need research. The skill builds a research step for the second group into the workflow.
3. Enrichment needs
Job titles to target, contact fields for your channel, and the research fields your template needs.
4. The message template and personalization
Personalization types Ben says work:
- Name-drop a second decision maker: “Wasn’t sure if you or {second decision maker} was the right one to contact for this.” He thinks it signals that you did your research.
- Name a recognizable customer in their niche: “We’ve been helping sales teams of fintech companies like {recognizable company}…”, with an offer to send that customer’s LinkedIn post about working with you.
- Mention something specific from their LinkedIn, website or a case study.
- Short, lowercase subject lines, which look like an email from a colleague.
He recommends the lavender.ai blog for cold-email templates. Give the skill two templates to build an A/B test into the output. Expect to iterate on messages before one gets replies.
The outbound skills in the sales plugin
The August sales video ships three outbound skills alongside the call-cycle skills in Ben AI’s sales skills:
- Lead generation (full SDR/BDR workflow in one skill). Give it your ICP (Ben’s came from his second brain), and it picks a source: Google Maps, Sales Navigator, Vibe Prospecting, Apollo or a directory. For his ICP (“SMBs that want to adopt AI”) it queried Vibe Prospecting for intent topics such as AI strategy and AI automation. Ben’s reported funnel: 246 prospects, 45 with verified emails, 20 unique companies. It then researches each one and outputs a prioritized sheet with a personalization toolkit.
- LinkedIn post engager. Scrapes everyone who commented on or reacted to a set of LinkedIn posts (yours, a competitor’s, or people in your space) via Apify, researches and deduplicates them, qualifies against the ICP, and finds emails. Ben’s run on his last five posts: 297 unique engagers, 129 qualified leads with verified emails. He says this warmer method has produced “quite a few closes” and is the one his team uses most. The sheet keeps the post each person engaged with, for personalization.
- CRM prospect mining. Goes through lost-deal columns (no-show, churned, stalled, lost), filters by ICP, reads the email and call-transcript history, researches each lead, and outputs a prioritized sheet with why each deal was lost and how to re-open it. Ben’s run surfaced 31 leads. Connectors: CRM, Apify, and optionally the transcription tool and email.
Try It
- Write down one ICP and one offer. One skill per pair.
- Pick the source from the table above and install the connector. For a local-business target, start with Apify’s Google Maps scraper.
- List your qualification criteria and mark which ones need a research step.
- Write the outreach template yourself with 3–4 variables (second decision maker, recognizable customer, specific detail). Use a short lowercase subject line. Add a second template if you want an A/B test.
- Run the flow once interactively for a small batch, check every row of the sheet, then tell Claude to “create a skill out of this workflow”.
- Upload the CSV to your sending tool. Ben uses Instantly. Schedule the skill only after a few manual runs look right.
- Build the warm variant next: a post-engager skill on your own LinkedIn posts, and a CRM-mining pass on last year’s lost deals.
Agency angle. For a dental-marketing agency, the local-business row is the one that matters: practices are on Google Maps, so an Apify Maps scrape plus a research step (for example, “has no online booking” or “fewer than N reviews”) gives a qualification filter that a database can’t. The CRM-mining skill applies directly to an agency’s own lost-proposal pipeline.
Implementation
Tool/Service: Claude Cowork or any skill-capable Claude surface; free skill from Ben’s Notion page (link in the video description). Setup: Upload the skill (Customize → Skills → upload), run it, answer the interview, approve the plan, review the sheet, then save the result as a skill. Cost: The skill is free. Apify, Vibe Prospecting, Anymail Finder and Instantly are paid per use or by subscription; the videos give no prices. Integration notes: Output is Google Sheet plus CSV. Sending happens in your outreach platform. Can run as a scheduled task.
Open Questions
- No performance numbers. “Performing really well” and “quite a few closes” are not backed by reply or close rates.
- Funnel numbers are Ben’s own runs (246 → 45 → 20; 297 → 129; 31 lost deals). Not verified.
- Tool names are auto-caption reconstructions. “Anymail Finder” and “lemlist” are inferred from the captions “any mailinder” and “Lamb list”.
- LinkedIn scraping terms of service. Neither video addresses LinkedIn’s restrictions on scraping engagers or profiles.
- Deliverability. Neither video covers domain warm-up or sending limits; see Clay + Claude Code for one creator’s numbers.
Related
- Ben AI’s Sales Skills — the call-cycle half of the same plugin
- Cowork + Apify Scraping Recipe — Eliot Prince’s Apify plus Vibe Prospecting flow, step by step
- The LinkedIn-Signal Outbound Agent — signal-based outbound with waterfall enrichment
- Clay + Claude Code — the Clay alternative, with credit costs and sending limits
- AutoResearch for Cold Outbound — an experiment loop for message testing
- Marketing Skills Bundle (Corey Haines) — its
/prospectingskill uses a similar ICP → qualify → score flow - Scheduled Tasks — running a finished skill on a schedule