Latest / AI Fire Daily / #137 Robin: This Free Claude Skill Can Find 100 Leads, Enrich Them, and Write Personalized Cold Emails

Transcript

#137 Robin: This Free Claude Skill Can Find 100 Leads, Enrich Them, and Write Personalized Cold Emails · AI Fire Daily

• 0:00
  Listen to this. What if you could generate 100
• 0:03
  qualified leads, find their emails, and write
• 0:06
  personalized copy using just one AI skill? That
• 0:10
  is exactly what we are unpacking today. Welcome
• 0:12
  to this deep dive. I'm really glad you were here.
• 0:16
  If you have ever, you know, spent a Tuesday afternoon
• 0:19
  staring blankly at a massive spreadsheet of dead
• 0:21
  -end leads, you know the struggle. Oh, yeah.
• 0:25
  Outbound sales can be incredibly soul -crushing.
• 0:29
  They really can. But today we have a mission.
• 0:32
  We are exploring a comprehensive AI lead generation
• 0:35
  workflow and It is built entirely around a free
• 0:39
  downloadable cloud skill Yeah, it completely
• 0:43
  changes the game. Honestly, I really does. So
• 0:46
  here is our roadmap for today First we will explore
• 0:50
  the core concept of AI driven outreach Then we
• 0:54
  will examine a precise four -step framework for
• 0:57
  sourcing and messaging And finally, we will show
• 1:00
  you how to automate this entire machine. Right,
• 1:02
  because you can build a repeatable process very
• 1:06
  quickly. Like the efficiency here is just incredible.
• 1:09
  Instead of you manually hunting down prospects,
• 1:12
  Claude just takes over. It handles the entire
• 1:14
  grueling pipeline. It finds your prospects. It
• 1:17
  checks if they actually fit your business model.
• 1:19
  Then it researches their specific contact details.
• 1:21
  And finally, it prepares personalized outreach
• 1:24
  messages for every single one. And I think the
• 1:27
  best part is really the barrier to entry. You
• 1:29
  can start this with a completely free template.
• 1:31
  You just download it and upload it directly into
• 1:34
  Claude. But we need to understand the philosophy
• 1:36
  behind it first. Yeah, that is a crucial point.
• 1:40
  People always ask for the absolute best lead
• 1:42
  source. But the best lead source depends entirely
• 1:45
  on your specific business. Makes sense. There
• 1:48
  is no universal answer. I mean, if you target
• 1:50
  local restaurants, Google Maps is perfect. But
• 1:53
  if you sell to SaaS companies, LinkedIn is just
• 1:55
  much better. So this Claude skill basically acts
• 1:59
  as a custom workflow creator. It is not a rigid
• 2:02
  tool forcing you into one specific path. It actually
• 2:05
  adapts to your specific service and target customers.
• 2:09
  Exactly. It's kind of like stacking Lego blocks
• 2:12
  of data. Oh, I like that. Yeah, you aren't buying
• 2:14
  a pre -built toy. You take a scraping block.
• 2:16
  you attach an enrichment block, then you just
• 2:19
  snap on a personalization block. You build exactly
• 2:22
  the vehicle your specific business needs. I love
• 2:26
  that analogy. It perfectly describes the modular
• 2:28
  nature of this whole system. Let us look at a
• 2:31
  specific scenario from our sources. Okay. Imagine
• 2:33
  an AI coaching program for business owners. They
• 2:36
  could run three totally different outreach campaigns
• 2:39
  using this exact same skill. Let's break those
• 2:42
  three examples down. Okay. The first campaign
• 2:44
  targets FinTech sales leaders. Right. So it finds
• 2:48
  sales managers at FinTech unicorns. Then it uses
• 2:51
  relevant client success stories to personalize
• 2:54
  the outreach. It basically proves you understand
• 2:56
  their specific industry. And then the second
• 2:58
  campaign targets SaaS companies that are actively
• 3:01
  hiring salespeople. Now this one is fascinating.
• 3:04
  Targeting companies actively hiring sales roles
• 3:06
  is highly strategic. It is a massive buying signal.
• 3:10
  It suggests that they are ready to invest in
• 3:12
  sales training right now. Yeah. And the third
• 3:15
  campaign targets LinkedIn post -engagers. Right.
• 3:18
  This campaign focuses on warm interactions. It
• 3:20
  finds people who recently liked or commented
• 3:22
  on your LinkedIn posts. Then it cross -references
• 3:25
  them against your ideal customer profile. So
• 3:27
  the basic workflow is identical for all three
• 3:29
  of those. You choose how many companies you want.
• 3:32
  Claude creates a comprehensive plan. You approve
• 3:35
  that plan. And then Claude collects and qualifies
• 3:38
  the leads. It finds their contact details. Then
• 3:41
  it writes the personalized messages. And finally,
• 3:44
  it exports everything directly to its spreadsheet.
• 3:46
  You can upload that right into an outreach platform.
• 3:49
  A tool like Instantly works perfectly for this.
• 3:52
  The results from this specific example are actually
• 3:55
  very revealing. The SaaS hiring campaign performed
• 3:58
  noticeably better than the Fintech one. I think
• 4:00
  there is a profound realization here. The insight
• 4:03
  isn't just about SaaS versus Fintech. It is about
• 4:06
  realizing the difference between an event and
• 4:09
  an identity. Oh, wow. That is a brilliant way
• 4:12
  to phrase it. Right. Because an identity is static.
• 4:16
  Being a fintech unicorn is just an identity.
• 4:19
  But hiring five new sales reps, that is a specific
• 4:23
  event. Events trigger new budgets. Exactly. That
• 4:28
  is such a useful signal. It tells you exactly
• 4:30
  where to put your effort. You should always start
• 4:32
  with two or three different lead sources. You
• 4:36
  must treat these campaigns as isolated experiments.
• 4:39
  I do have to push back a little here, though.
• 4:41
  Does this ease of use just create more digital
• 4:44
  noise? Like, are we just generating more spam
• 4:47
  if we aren't managing this carefully? That is
• 4:49
  a very valid concern. If you just blast generic
• 4:52
  messages, yes, it absolutely becomes spam. That
• 4:56
  is why the experiment mindset is so critical.
• 4:58
  You use the tool to test different audiences
• 5:00
  quickly. Then you focus only on the ones bringing
• 5:03
  in the best replies. Quality always beats sheer
• 5:06
  volume. Which makes sense. So what makes a hiring
• 5:10
  announcement such a strong buying signal? Well,
• 5:13
  when a company expands its sales team, they need
• 5:16
  onboarding. They need immediate performance from
• 5:18
  those new hires. That urgency almost always means
• 5:21
  they have allocated capital for external coaching.
• 5:24
  So hiring signals indicate they have immediate
• 5:27
  budget for training. Exactly. It completely changes
• 5:30
  the conversation. To run these experiments, we
• 5:33
  first have to figure out exactly where these
• 5:35
  ideal leads actually live on the internet. Yeah.
• 5:39
  Which brings us to step one of the four -step
• 5:41
  framework. Choosing where to find your leads
• 5:44
  is the most important decision. You want to find
• 5:47
  prospects where they already exist. Different
• 5:50
  niches require completely different tools. Let
• 5:52
  us examine the tech stack we are using here.
• 5:55
  We are dropping some technical names, so let's
• 5:57
  clarify them. For local businesses, you have
• 6:00
  a few options. Google Maps via Appify is highly
• 6:02
  recommended. Right. Think of Appify as the bridge.
• 6:06
  It connects Claude's brain to the raw data of
• 6:09
  the internet. Okay. If you target tech, SaaS,
• 6:12
  or B2B companies, the landscape changes. You
• 6:15
  want to use LinkedIn or Apollo. Lemlist is another
• 6:18
  great option. Yeah. You can use LinkedIn sales
• 6:20
  navigator scrapers via Appify for this. But what
• 6:24
  if you were looking for those crucial buying
• 6:25
  signals we just mentioned? Vibe prospecting is
• 6:28
  excellent for that. It really is. It tracks hiring
• 6:31
  trends and funding rounds. It also flags technology
• 6:34
  changes and new leadership appointments. For
• 6:37
  influencers and creators, social media platforms
• 6:40
  are best. Smaller, specific niches require industry
• 6:44
  directories. Listing websites or Google results
• 6:47
  via Firecrawl are ideal there. Think of Firecrawl
• 6:50
  as the eyes of your AI. It scans the web efficiently.
• 6:55
  And the primary recommendation here is very powerful.
• 6:58
  You should connect both Apify and Firecrawl to
• 7:01
  Claude. Together, they collect data from a huge
• 7:04
  range of public websites. This covers most common
• 7:07
  lead generation needs. We should define that
• 7:09
  jargon really quickly. What exactly is a scraping
• 7:13
  tool? It is software that extracts website data
• 7:17
  automatically so you don't have to. That makes
• 7:19
  perfect sense. So let's picture a practical campaign
• 7:23
  setup. Claude might recommend a niche business
• 7:26
  directory as a lead source. You use Firecrawl
• 7:29
  as your scraping tool. You target, say, 100 leads
• 7:32
  per run. Right. And once you choose that source,
• 7:35
  Claude builds the scraping process. But scraping
• 7:38
  is only the beginning. Step two is defining your
• 7:41
  lead qualification criteria. Getting 1 ,000 leads
• 7:45
  sounds incredibly appealing at first, but how
• 7:48
  many of them would actually buy your service?
• 7:50
  I'll be honest, I... I still wrestle with prompt
• 7:52
  drift myself when staring at massive lists. You
• 7:56
  just start wanting to email everyone. Oh, we
• 7:58
  all do. It's so tempting. That is why lead qualification
• 8:01
  matters so much. Some databases let you filter
• 8:04
  companies by basic metrics. You can sort by location,
• 8:07
  employee count, or revenue. But those basic filters
• 8:10
  often miss the critical details. They miss the
• 8:12
  nuances that really matter for your specific
• 8:14
  business. Right. Let us return to our fintech
• 8:17
  example. Say you only want companies with sales
• 8:20
  teams larger than 25 people. Most standard lead
• 8:24
  databases do not provide that exact headcount.
• 8:27
  This is where the workflow becomes truly intelligent.
• 8:30
  You ask Claude to add an extra research step
• 8:32
  after scraping. Exactly. It will independently
• 8:35
  investigate each company. Because Claude understands
• 8:38
  context, it doesn't just look for the number
• 8:40
  25. It actually executes human -like reasoning
• 8:43
  at scale. This helps you avoid wasting time contacting
• 8:47
  the wrong people. This structured qualification
• 8:50
  solves the volume problem entirely. A smaller
• 8:53
  list of 100 highly relevant prospects is always
• 8:56
  better. It's far more valuable than a thousand
• 8:59
  random contacts. Quality data is the entire foundation
• 9:03
  of this system. But you're telling me a free
• 9:05
  AI tool can cross -reference employee headcounts
• 9:09
  completely unassisted. How does Claude actually
• 9:12
  verify something as specific as team size? It
• 9:15
  navigates to the company's About Us or team pages.
• 9:18
  It parses the structure of the team. It cross
• 9:21
  -references employee directories or LinkedIn
• 9:23
  company data to deduce specific roles. It scans
• 9:27
  company pages to confirm exact headcount before
• 9:30
  approving. Precisely. It acts like a junior researcher
• 9:33
  doing manual diligence. Once we filter the noise
• 9:36
  and found those 100 perfect fits, We need to
• 9:39
  figure out exactly who to contact and what to
• 9:42
  say without sounding like a robot. This is step
• 9:44
  three. We call this process lead enrichment.
• 9:47
  It usually involves three distinct pillars. The
• 9:50
  first pillar is finding the actual decision makers.
• 9:53
  You need the job titles most likely to need your
• 9:55
  service because a list of companies doesn't do
• 9:58
  us any good if we don't know who's inbox to land
• 10:01
  in. Right. For sales training, you want a VP
• 10:04
  of sales. A head of sales or a sales director
• 10:07
  also works perfectly. And then the second pillar
• 10:09
  is finding the contact details. You need verified
• 10:12
  work emails or LinkedIn profiles. An email finder
• 10:15
  is a great option for email outreach. It specializes
• 10:19
  in finding verified work email addresses. If
• 10:21
  you haven't connected an email finder, Claude
• 10:23
  can actually just recommend one. The third pillar
• 10:26
  is personalization data. You must research useful
• 10:29
  details about their company. You look for recent
• 10:31
  activities or pressing business challenges. The
• 10:34
  goal here is actually quite simple. You must
• 10:37
  give Claude enough accurate information first.
• 10:40
  It needs this data to identify the right person.
• 10:43
  Then it can write a message that feels genuinely
• 10:45
  relevant. That leads directly into step four,
• 10:48
  writing personalized outreach messages. The source
• 10:51
  material has a very crucial warning here. You
• 10:53
  should not let Claude write your cold email from
• 10:56
  scratch. Oh yeah. It is a massive mistake. You
• 10:59
  will usually get much better results with a template.
• 11:02
  You provide a strong proven email structure first.
• 11:05
  Then you let Claude fill in the specific details
• 11:08
  using its research. There are four specific personalization
• 11:11
  tactics worth testing here. The first tactic
• 11:13
  is mentioning another decision maker by name.
• 11:16
  You might write, wasn't sure if you or Sarah
• 11:18
  was the right person. You must always use a real
• 11:22
  verified name for this, though. The second tactic
• 11:24
  is using social proof. You mentioned a recognizable
• 11:28
  client you have successfully helped. Ideally,
• 11:30
  you back this up with a case study. A direct
• 11:33
  testimonial makes the social proof much stronger.
• 11:35
  The third tactic is referencing something very
• 11:38
  specific. A recent LinkedIn post gives you a
• 11:41
  natural reason to connect. A company announcement
• 11:44
  or a recent funding round works well too. And
• 11:47
  the fourth tactic is about the subject line itself.
• 11:50
  Keep subject lines short and entirely lowercase.
• 11:53
  For example, you might write, quick question
• 11:55
  about your sales team? The psychology behind
• 11:58
  that short lowercase subject line is brilliant.
• 12:01
  It feels like an internal casual memo from a
• 12:04
  colleague. It does not look like a polished corporate
• 12:07
  marketing blast. It bypasses that initial mental
• 12:10
  spam filter we all have. It creates a sense of
• 12:13
  casual urgency. You can look at the Lavender
• 12:15
  .ai blog for more examples. They have excellent
• 12:18
  outreach templates and data -driven writing tips.
• 12:21
  One final tip for this section is incredibly
• 12:24
  important. You must test your messaging rigorously.
• 12:27
  Yeah, you can give Claude two completely different
• 12:30
  email templates. Then you ask it to prepare a
• 12:32
  structured A -B test. You compare the reply rates
• 12:35
  later. This shows you exactly which approach
• 12:37
  actually performs better. That is how you gradually
• 12:40
  improve your AI campaigns over time. You stop
• 12:42
  guessing and start relying on actual data. Data
• 12:45
  is the only truth in cold outreach. So why does
• 12:49
  letting AI write an email from scratch fail so
• 12:51
  often? Language models tend to use overly formal,
• 12:55
  flowery language by default. They lack the punchy
• 12:57
  conversational brevity required for effective
• 12:59
  cold sales. AI writing from scratch usually sounds
• 13:02
  too generic and robotic. Exactly. The template
• 13:05
  keeps the AI strictly grounded. So, yeah, the
• 13:11
  template is everything. Now that we have our
• 13:13
  qualified data and a perfectly human sounding
• 13:15
  message, the final piece is taking our hands
• 13:18
  off the wheel and letting the system run. This
• 13:22
  is where all the setup finally pays off. Once
• 13:24
  you complete the four steps, Claude builds the
• 13:26
  plan. It puts everything together into a comprehensive
• 13:29
  lead generation workflow. You will see exactly
• 13:32
  how the workflow will run first. You review this
• 13:34
  before any actual data is ever collected. It
• 13:37
  outlines the lead source and the scraping tool.
• 13:40
  It shows you exactly which connector it will
• 13:42
  use. It details the qualification and enrichment
• 13:45
  rules clearly. It shows how it will filter leads
• 13:48
  and find decision makers. It also details the
• 13:51
  exact personalization variables it will use.
• 13:54
  It shows which research details will fill in
• 13:56
  your outreach template. You review this plan
• 13:58
  carefully. You make any necessary changes. Then
• 14:01
  you just let Claude run the entire process. The
• 14:03
  output is incredibly clean and organized. You
• 14:06
  get a Google Sheet and a CSV file. It contains
• 14:09
  your qualified leads and their company information.
• 14:12
  It has the contact details and the personalized
• 14:14
  outreach messages. And you can upload that CSV
• 14:17
  directly to an outreach platform. Instantly is
• 14:20
  perfect for this. Your campaign is practically
• 14:23
  ready to launch immediately. Now we arrive at
• 14:26
  the absolute best part. You can save this entire
• 14:29
  setup as a reusable Claude skill. Yes. Once you
• 14:32
  test the workflow and confirm it works, you lock
• 14:34
  it. You use one specific prompt to automate the
• 14:37
  whole thing. The prompt is very straightforward.
• 14:40
  You tell Claude to turn this lead generation
• 14:42
  workflow into a reusable skill. You instruct
• 14:45
  it to keep the exact same lead sources. You lock
• 14:48
  in the qualification rules and the enrichment
• 14:50
  steps. You also lock in this specific outreach
• 14:52
  template. This allows you to run it again easily
• 14:55
  for new leads. Claude creates a customized skill
• 14:58
  you can save permanently. You can use it again
• 15:01
  whenever you need fresh prospects. You can take
• 15:03
  this automation even further if you want. You
• 15:05
  can set it up as a scheduled recurring task.
• 15:08
  You just need a supported environment to host
• 15:10
  it. That means the workflow runs regularly on
• 15:13
  its own. It collects new leads and researches
• 15:15
  their details automatically. It prepares fresh
• 15:18
  outreach messages without any daily input from
• 15:20
  you. Wow! Imagine scaling this to thousands of
• 15:24
  personalized touch points while you sleep. Right,
• 15:26
  and that scale is incredibly powerful, but we
• 15:28
  must address the final piece of advice here.
• 15:31
  Yes, you must run it manually a few times first.
• 15:34
  This is a non -negotiable step for long -term
• 15:36
  success. You have to check that the leads are
• 15:38
  actually relevant. You must verify the contact
• 15:41
  details are consistently accurate. Most importantly,
• 15:45
  you must ensure the generated messages still
• 15:47
  sound natural. You do this before scheduling
• 15:49
  the fully automated process. Once everything
• 15:51
  looks perfect, you have a repeatable system.
• 15:54
  You have a lead generation engine ready to scale
• 15:56
  safely. What is the biggest risk of scheduling
• 16:00
  these tasks without manual testing first? If
• 16:04
  the AI hallucinates a data point, it scales that
• 16:07
  error. You could end up insulting a hundred executives
• 16:10
  with wildly inaccurate claims. Manual testing
• 16:13
  catches bad data before automation ruins your
• 16:16
  reputation. Exactly. Trust the system, but verify
• 16:21
  the output first. Let us step back and look at
• 16:24
  the bigger picture here. AI is clearly not here
• 16:27
  to replace human strategy. It is not replacing
• 16:30
  the art of true relationship building either.
• 16:33
  It is here to scale the grueling execution of
• 16:36
  outreach. The manual research phase has always
• 16:39
  been the hardest part. By starting small with
• 16:41
  two or three lead sources, you learn. You give
• 16:44
  the AI strict parameters and test your templates.
• 16:47
  You build a repeatable system rather than shooting
• 16:50
  in the dark. You let the machine do the heavy
• 16:52
  lifting of data collection. You let it handle
• 16:55
  the tedious cross -referencing and verification.
• 16:58
  But you retain total control over the strategic
• 17:01
  direction. You dictate the tone and the overarching
• 17:04
  message. It elevates your role from a data scraper
• 17:07
  to a strategist. You spend your time talking
• 17:09
  to the qualified leads. You are not spending
• 17:12
  hours trying to find their email addresses. This
• 17:15
  shift changes the entire economics of outbound
• 17:17
  sales. A single person can now operate like an
• 17:21
  entire SDR team. But that brings us to a rather
• 17:24
  fascinating paradox. We are using highly advanced
• 17:27
  AI to sound more perfectly human. We are designing
• 17:31
  lowercase subject lines to mimic casual human
• 17:35
  imperfection. It is deeply ironic when you think
• 17:39
  about it. It really is. As these highly personalized
• 17:42
  automated AI outreach tools become the standard
• 17:45
  for every business, how will we distinguish genuine
• 17:48
  human connection from a perfectly tuned Claude
• 17:51
  skill? What does the inbox of the future look
• 17:54
  like when everyone is this good at cold emailing?
• 17:58
  That is a wild thought to leave on. Thank you
• 18:00
  for joining us on this deep dive today. Keep
• 18:03
  building, keep testing, and we will see you next
• 18:05
  time.