Automate Website Leads to Voice Demo & Scheduling: A Founder’s Story

How Alex’s “Leaky Funnel” Turned Into an AI Booking Machine

Alex stared at the CRM again. Dozens of new leads had come in from the website over the weekend, but only two had actually booked a call. The rest were stuck in limbo, tagged as “New,” waiting for a human follow-up that never arrived on time.

Like many founders, Alex had a familiar problem. The website form worked. Ads were driving traffic. People were interested. Yet the manual process of qualifying leads, calling them, and scheduling demos was slow, inconsistent, and painfully human-dependent.

Some leads never picked up. Some got a rushed call with no context. Others slipped through the cracks entirely. Even worse, every update to the CRM had to be done by hand, which meant it rarely reflected reality.

Alex wanted something different: a system that would notice every new lead, understand their business, call them with a personalized pitch, and log everything automatically. No more “I’ll get to it later.” No more missed opportunities.

That is when Alex discovered an n8n workflow template that promised exactly that – an AI booking agent powered by n8n, Notion, OpenRouter, and Vapi.

The Vision: Turning a Website Form Into an AI Booking Agent

Instead of treating the website form as a passive inbox, Alex wanted it to act like a proactive AI sales assistant. The idea was simple but powerful:

• Every time a visitor submitted a form, the system would create a new lead in Notion.
• An n8n workflow would detect that lead, research the prospect’s website, and summarize their business using AI.
• An AI voice assistant in Vapi would call the lead, speak intelligently about their business, and try to schedule a meeting.
• Once the call ended, the system would store the outcome, notes, and recording back in Notion.

In other words, Alex wanted a fully automated AI booking agent that could qualify, call, and follow up with leads, all without a human touching the process.

Gathering the Tools: What Alex Needed to Get Started

Before anything could work, Alex had to assemble the stack. The template made it clear what was required, and it turned out to be surprisingly straightforward:

• n8n account for automation, either cloud or self-hosted.
• Notion account, even the Free plan, to act as the CRM and central lead database.
• OpenRouter API key from openrouter.ai to power the AI text analysis with GPT-3.5.
• Vapi account from vapi.ai to handle AI voice calls.

Inside Vapi, Alex created an AI assistant and assigned it a phone number. The system then generated three critical pieces of information:

• Vapi API key
• Assistant ID
• Phone Number ID

Those credentials would later plug directly into the n8n workflow.

On the CRM side, Alex duplicated the provided Notion template into the workspace. That template came pre-structured with the fields the workflow needed, which meant fewer chances to misconfigure things.

The First Breakthrough: Seeing the Automation Flow in Action

Once the accounts were ready, Alex opened the n8n template. Instead of a simple one-step zap, this was a complete automation journey, from new lead to completed call.

Here is how Alex’s new AI booking agent would behave behind the scenes.

From Form Submission to AI Voice Call

1. New Lead Detection
   Every time a visitor filled out the website form, a new record appeared in the Notion database with Status = “New”. Alex did not have to touch anything. That “New” label was the signal that a fresh lead had arrived.
2. n8n Trigger on Notion
   In n8n, a Notion Trigger node checked for new leads every minute. When it spotted a record with Status set to “New,” the workflow kicked in.
3. Website Content Fetch
   The workflow pulled in the lead’s website content from the URL they submitted. During testing, Alex used the built-in mock mode to simulate this step without hitting real sites, which made safe experimentation possible.
4. AI Business Analysis with OpenRouter
   n8n then sent the website content to GPT-3.5 via the OpenRouter API. The AI analyzed the business, summarized what the company did, and even extracted an interesting fact. This was the magic ingredient that made calls feel tailored instead of generic.
5. Update Notion with Insights
   That AI-generated summary was written back into the same Notion record, filling the “Business Analysis” field. Now each lead entry contained not just raw contact info, but a concise snapshot of their business.
6. Vapi AI Call
   Using the Vapi API, the workflow instructed the AI assistant to call the lead’s phone number. The assistant used the business analysis to open the conversation in a relevant, human-sounding way, rather than with a canned script.
7. Webhook on Call Completion
   When the call ended, Vapi sent a webhook back to n8n. That webhook was the trigger for the second part of the workflow, which handled the outcome.
8. AI Call Summary
   The call’s content was passed again through AI to produce a clean summary and actionable notes. If the lead did not answer, the workflow could reuse an existing summary or log that outcome appropriately.
9. Final Notion Update
   Finally, the Notion record was updated with:
   • Call summary
   • Recording link
   • Call outcome, such as “Meeting Scheduled” or “No Answer”

   What used to be a vague “New” entry was now a fully documented interaction.

Inside the Workflow: The Two Key Parts Alex Activated

As Alex explored the template, it became clear that the automation was divided into two cooperating workflows, each responsible for a different stage of the journey.

Part 1 – From New Notion Lead to AI Voice Call

The first workflow handled everything from detecting a new lead to making the call. In Alex’s words, it was like a virtual SDR that never slept.

• It watched the Notion database for leads where Status was exactly “New”.
• It fetched the lead’s website and cleaned up the content so the AI had something usable to analyze.
• It sent that content to GPT-3.5 via OpenRouter to create a business summary and an interesting fact.
• It stored the AI analysis back into the Notion record so the CRM was enriched automatically.
• It triggered a Vapi AI voice call using the Phone Number ID and Assistant ID, passing along the analysis so the assistant could sound informed and relevant.

After setting this up, Alex realized that the hardest part of lead qualification was now handled by automation.

Part 2 – Webhook Handler for Call Results

The second workflow lived in the background, waiting for Vapi to report back after each call.

• It received the webhook from Vapi whenever a call ended.
• It used the Vapi API to fetch detailed call results, including transcript or summary data.
• It generated an AI call summary if needed, or reused existing information when the call was not answered.
• It updated the Notion record with:
  • Call outcome (for example, “Meeting Scheduled” or “No Answer”)
  • Summary and notes
  • Recording link

This was the moment Alex realized something important. The team would never again wonder, “Did we call this lead?” The history was always there, cleanly logged, a few seconds after each interaction.

The Setup Hurdle: Getting the Configuration Right

Of course, no automation story is complete without a bit of configuration tension. For Alex, that came in the form of environment variables and webhook URLs.

The template required several placeholders to be replaced before anything would run:

• YOUR_VAPI_API_KEY
• YOUR_VAPI_ASSISTANT_ID
• YOUR_VAPI_PHONE_NUMBER_ID

Alex carefully swapped each placeholder with the actual values from the Vapi dashboard. A single typo here could break the flow, so this was double-checked.

Next, the Notion database needed the correct fields so n8n could read and write data reliably. The required structure looked like this:

• Name – The lead’s name.
• Phone – The number Vapi would call.
• Website URL – Used to fetch and analyze the business site.
• Status (select) – Values like “New,” “Meeting Scheduled,” or “No Answer.”
• Business Analysis (rich text) – Where the AI summary would be stored.

Finally, Alex had to connect the dots between Vapi and n8n for the second workflow. In the Vapi assistant dashboard, the webhook URL from Part 2 of the n8n workflow was pasted into the correct field. That ensured every completed call would notify n8n automatically.

To avoid burning through real calls during testing, Alex enabled the mock data mode included in the template. This allowed full end-to-end checks of the logic without dialing actual phone numbers.

The Turning Point: From Missed Leads to Automated Conversations

With everything configured, Alex flipped the switch by activating the Part 1 workflow in n8n. The Notion Trigger started monitoring leads in the background.

Within minutes, a new test lead appeared in Notion with Status set to “New.” The workflow caught it, fetched the website, generated the analysis, and logged the insights. Then Vapi placed the call.

The AI assistant greeted the lead using their name, referenced their business correctly, and asked thoughtful questions. It did not sound like a cold script, it sounded like someone who had done their homework.

After the call, Alex refreshed Notion. The record now contained:

• A clear call summary
• Notes on what the lead cared about
• The call outcome, including whether a meeting had been scheduled
• A link to the recording for later review

The entire process, from form submission to follow-up and logging, had happened automatically.

The Resolution: What Changed for Alex’s Team

Over the next few weeks, the impact was obvious:

• No new lead sat untouched in Notion. The AI booking agent reached out quickly and consistently.
• Every call felt personalized, thanks to the AI business analysis powered by OpenRouter and GPT-3.5.
• The sales team had clean, structured notes in Notion before they even joined a scheduled meeting.
• Missed calls were no longer black boxes. They were logged with outcomes like “No Answer,” so follow-up strategies could be adjusted.

Most importantly, Alex stopped worrying about whether anyone had remembered to call that promising lead from Friday afternoon. The system handled it, every time.

Set Up Your Own AI Booking Agent With n8n

If Alex’s story feels familiar, you might be in the same place: plenty of leads, not enough time, and too much manual follow-up.

This n8n workflow template gives you a practical way to:

• Automate lead qualification directly from your website form.
• Use an AI voice assistant to call leads with personalized context.
• Keep your Notion CRM updated with business insights, call summaries, and outcomes.
• Test safely using mock data mode before going live with real calls.

The steps are clear:

1. Create or log into your n8n, Notion, OpenRouter, and Vapi accounts.
2. Set up your AI assistant and phone number in Vapi, then copy your API key, Assistant ID, and Phone Number ID.
3. Duplicate the Notion template and ensure all required fields are in place.
4. Import the n8n workflow template, replace the placeholders with your credentials, and configure the Vapi webhook URL for Part 2.
5. Activate the Part 1 workflow to start monitoring and calling new leads.

Next Step: Try the Template

You do not have to rebuild Alex’s system from scratch. The template already connects n8n, Notion, OpenRouter, and Vapi into a ready-to-customize AI booking agent.

Set up your accounts, duplicate the Notion database, and plug in your credentials. If you want help tailoring the flow to your specific sales process, you can always reach out for integration and optimization support.