Build an AI Lead Qualifier That Never Sleeps

Your Lead Inbox is a Crime Scene. Let’s Investigate.

It’s 9 AM. You open your inbox, feeling hopeful. A notification glows: “You have 15 new contact form submissions!”

Your heart skips a beat. Is this it? The big client? The game-changing partnership?

You click. And reality hits you like a sack of wet spam.

Lead 1: “hi can u send me info” (From a student with a Hotmail address).
Lead 2: A competitor trying to see your pricing.
Lead 3: Someone selling you SEO services from a country you can’t find on a map.
…and so on, for 14 soul-crushing entries.

But wait. Buried at the bottom, Lead #15 is different. It’s from a Director at a Fortune 500 company, with a specific request and a stated budget. It’s a whale. And it’s been sitting there for 11 hours while you were sleeping.

This, my friends, is Lead Inbox Hell. It’s a place where golden opportunities go to die, suffocated by an avalanche of digital noise. Every hour you spend manually sifting through this mess is an hour you’re not talking to that whale. We’re going to fix that. Today.

Why This Matters

Let’s be blunt. Manually qualifying leads is a terrible use of a smart person’s time. It’s a job for a robot. Or better yet, an AI intern who works 24/7, never complains, and costs less than your morning coffee.

Speed to Lead is Everything: Studies show that contacting a new lead within 5 minutes increases your chances of qualifying them by a factor of nine. Your AI intern can flag a hot lead in seconds.
Focus on Revenue, Not Rubbish: Your sales team (or you, if you’re a one-person army) should spend their time closing deals, not acting as a human spam filter. This system lets them focus exclusively on pre-qualified, high-potential conversations.
Never Miss an Opportunity: The AI works while you sleep, during holidays, and even when you’re binge-watching that new sci-fi show. A million-dollar lead can come in at 3 AM on a Sunday. This system will catch it.

We are not just building a fancy notification system. We are building a money-making machine that protects your most valuable asset: your time.

What This Workflow Actually Is

Think of it as a tiny, automated factory for processing leads. It has three main stations:

The Loading Dock (The Trigger): This is where the raw materials arrive. In our case, it’s a submission from your website’s contact form. We’ll use a special tool called a Webhook, which is like a digital doorbell that rings whenever a new form is submitted.
The Quality Control Inspector (The AI Brain): This is our AI model (we’ll use Anthropic’s Claude). It receives the lead’s message and acts like a hyper-efficient, senior-level consultant. It reads the message, understands the intent, checks it against your criteria (budget, company size, urgency), and stamps it with a grade.
The Sorting Chute (The Action): Based on the AI’s grade, the factory routes the lead. Hot leads get shot into a priority Slack channel with a loud notification. Warm leads might go into a Google Sheet for a weekly follow-up. Junk leads? They go straight to the digital incinerator (or are just ignored).

The whole thing is held together by an automation platform like Make.com (formerly Integromat), which is the factory’s conveyor belt system, moving the data from one station to the next.

Prerequisites (The Honest-to-Goodness Shopping List)

No magic here. You’ll need a few things. Don’t worry, most of this is free to start.

A Website Form That Can Send Webhooks: You need a way to get the lead data out of your website. Tools like Tally.so, Webflow, Carrd, or even WordPress plugins like Gravity Forms can do this easily. If you don’t have one, you can start with a free Tally.so form for this lesson.
A Make.com Account: This is our automation glue. The free plan is more than enough to build this entire system and run it 1,000 times a month.
An Anthropic (Claude) API Key: This is the AI brain. Go to the Anthropic console, sign up, and grab your API key. They give you some free credits to start, and after that, it costs fractions of a penny per lead. Seriously, you’ll probably spend $1-2 per month.
A Slack Workspace: This is where we’ll send our alerts for hot leads. A free Slack account is perfectly fine.

That’s it. No coding experience required. Just a willingness to click some buttons and copy-paste.

Step-by-Step Tutorial: Building Your AI Lead Sorter

Alright, class is in session. Open up Make.com and let’s build this factory.

Step 1: The Trigger – Create a Digital Mailbox (Webhook)

Our first step is to create a unique web address where your website form can send its data.

In a new Make.com scenario, click the big plus button and search for “Webhooks”.
Select “Custom webhook” as the trigger.
Click “Add”, give your webhook a name (like “Website Lead Catcher”), and click “Save”.
Make will generate a unique URL. COPY THIS URL. This is your digital mailbox. Don’t close this window yet! Make needs to “listen” to see what kind of data to expect.
Now, go to your website form builder (Tally, Webflow, etc.) and find the “Webhook” integration. Paste the URL you just copied.
Save your form, then go to the live form and submit a test entry. Fill it out with some sample data like a real lead would. For example: Name: “Bob Vance”, Email: “bob@vance-refrigeration.com”, Message: “We are a mid-size paper company looking to automate our supply chain. Our budget is around $25,000. We need a solution within the next quarter.”
Go back to Make.com. If everything worked, you’ll see a message that says “Successfully determined.” This means your webhook caught the test data and now understands the structure.

Step 2: The Brain – Ask Claude to Analyze the Lead

Now that we have the lead data, let’s send it to our AI inspector.

Click the plus icon to add another module to your scenario. Search for “Anthropic”.
Select the “Create a Message” action.
If this is your first time, you’ll need to add a connection. Click “Add”, and paste your Anthropic API Key in the box.
Now, we configure the AI’s instructions. This is the most important part!
For the Model, choose a fast and cheap one like `claude-3-haiku-20240307`.
In the Messages -> Text box, we will write our prompt. This is the SOP we’re giving our AI intern. You can map the data from your form directly into the prompt. Copy and paste this entire block:

You are an expert B2B lead qualification assistant. Your job is to analyze an incoming sales inquiry and determine its quality based on our criteria. Provide your analysis ONLY in a clean JSON format. Do not add any text before or after the JSON block.

Here is the lead's information:
Name: {{1.name}}
Email: {{1.email}}
Message: {{1.message}}

Analyze the message and determine the following:
1. lead_score: A score from 0 to 100, where 100 is a perfect lead. Base this on their budget, clarity of need, company size, and urgency.
2. category: Classify the lead into one of three categories: "Hot", "Warm", or "Junk".
3. summary: A one-sentence summary of the lead's request.
4. reason_for_category: A brief explanation for your classification.

Our ideal customer is a business (not a student or individual) with a clear problem, a stated budget over $5,000, and a timeline.

Return your output as a JSON object like this:
{
  "lead_score": 85,
  "category": "Hot",
  "summary": "A mid-size paper company wants to automate their supply chain with a $25,000 budget.",
  "reason_for_category": "Clear business need, stated budget, and defined timeline."
}

Why this prompt is so good: It tells the AI its role, gives it the raw data, defines the exact criteria for success, and most importantly, forces it to reply in a structured JSON format that our automation can easily understand.

Step 3: The Translator – Parse the AI’s Response

Claude will give us back a chunk of text that looks like JSON. We need to convert that text into usable data fields.

Add another module. Search for “JSON” and select “Parse JSON”.
In the JSON string field, map the output from the Anthropic module. It will be something like {{2.choices[].message.content}}.
You don’t need a Data Structure. Make will figure it out automatically when you run a test.

Step 4: The Traffic Cop – Route the Leads

Now we use a Router to send the lead down different paths based on its category.

Add another module. This time, search for “Router”. This will create a branching point in your scenario.
You will now see two paths branching from the router. Let’s configure the top path for “Hot” leads.
Click on the little wrench icon on the line connecting the router to the top path. This is the filter.
Set up the filter: Label it “Hot Leads”. For the condition, map the `category` field from the “Parse JSON” module. Set the condition to be Text operators: Equal to and type `Hot` in the value box.
Now, on that top path, add a “Slack” module and choose “Create a Message”.
Connect your Slack account. Choose a channel (e.g., `#sales-alerts`).
Now compose your message using the clean data from the “Parse JSON” module! Make it look nice and actionable. Here’s an example for the Text field:

:fire: NEW HOT LEAD :fire:

*Score:* {{3.lead_score}}/100
*Summary:* {{3.summary}}
*Reason:* {{3.reason_for_category}}

*Original Message:*
> Name: {{1.name}}
> Email: {{1.email}}
> Message: {{1.message}}

You can create another path from the Router for “Warm” leads that adds a row to a Google Sheet, and a default path that does nothing for “Junk”.

Finally, turn your scenario ON. Congratulations, your AI intern is now on duty.

Real Business Use Cases

Digital Agency: A web design agency uses this to instantly separate inquiries about “$50,000 redesign projects” from “Can you fix my nephew’s WordPress site for $50?”. The sales director gets a Slack alert and can call the big lead within minutes.
SaaS Company: A B2B SaaS founder filters demo requests. The AI identifies requests from large companies (e.g., anyone with a corporate email, mentions of “enterprise” or “multiple seats”) and routes them directly to the founder’s calendar booking link.
Real Estate Agent: An agent’s website gets dozens of inquiries. The AI analyzes the message for keywords like “pre-approved”, “looking to buy soon”, or “specific neighborhood” to flag motivated buyers and send a push notification straight to the agent’s phone.

Common Mistakes & Gotchas

Vague Prompting: If your prompt is lazy, your AI will be lazy. If you just say “Is this a good lead?”, you’ll get garbage back. Be specific. Give it criteria. Tell it EXACTLY how to format the output.
Not Testing with Bad Leads: Your system works great when you test it with a perfect lead. But what happens when someone types “asdfghjkl” into the form? Make sure you test junk leads to ensure they are categorized correctly and don’t accidentally fire off a HIGH ALERT.
Forgetting About Costs: The cost per lead is tiny, but it’s not zero. If your site gets hit with a spam bot that submits the form 10,000 times, you could get an unexpected bill. Set up a budget alert in your Anthropic account.
Parsing Errors: Sometimes, an AI might add a weird character or an extra sentence outside the JSON block, causing the “Parse JSON” module to fail. A more advanced trick is to use a text parser to extract just the content between the `{` and `}` before parsing.

How This Fits Into a Bigger Automation System

This lead qualifier is just the front door. It’s a powerful first step, but it’s not the end of the line. Think bigger.

Once a lead is flagged as “Hot,” our system could:

Automatically create a contact in your CRM (like HubSpot or Pipedrive).
Enrich the lead’s data using a tool like Clearbit to find their company size and LinkedIn profile.
Trigger another AI workflow to draft a personalized opening email based on their specific request. For example: “Hey Bob, saw you were looking to automate your supply chain for Vance Refrigeration. We have experience with mid-size distribution companies…”
Add them to a specific email sequence tailored to their needs.

Our simple sorter becomes the start of a fully automated “Inquiry-to-Proposal” pipeline, freeing up humans to do what they do best: build relationships and close deals.

What to Learn Next

Our AI intern is now a world-class receptionist, expertly sorting the good from the bad. But what if it could also handle the first conversation?

In the next lesson, we’re going to give our intern a promotion. We’ll teach it to take the information from a hot lead and automatically draft a personalized, non-robotic, and context-aware follow-up email. We’re moving from sorting mail to writing the first draft. Get ready to take back your inbox for good.

“,
“seo_tags”: “ai automation, lead qualification, make.com tutorial, business process automation, claude api, no-code ai”,
“suggested_category”: “AI Automation Courses