Author(s): Anna Jey

Originally published on Towards AI.

A step-by-step guide to using Make.com and Perplexity to automate your pre-meeting research and reclaim your mental energy.

We have all been there.

It is 1:58 PM. You have a Zoom call at 2:00 PM with a potential lead, a partner, or a candidate you have never met. You are frantically opening tabs (LinkedIn, their company website, their latest posts on X) trying to figure out who they are and what they care about right now.

You enter the meeting slightly out of breath, mentally scattered, and you end up regurgitating a generic “How are things in [City]?” icebreaker.

This is a waste of your mental energy. It is also completely avoidable.

As an automation consultant, I tell my clients a hard truth: Research is high-value work, but searching is low-value labor. You should spend your time reading the insights and forming a strategy, not clicking through Google results like a panicked intern.

In this guide, I am going to show you how to build a “Zero-Prep” Dossier System.

This isn’t just another chatbot wrapper. It is a live automation loop that watches your calendar, identifies external guests, sends an AI agent to scour the live web for recent news about them, and delivers a concise briefing document to your Slack or Notion 15 minutes before you speak.

Here is how to build it in 2026 without writing a single line of code.

The “Iron Man” Protocol: How It Works

Most people use AI incorrectly because they treat it like a search bar. They type a question, wait, and read. To get real leverage, you need to treat AI like a background process (a daemon, for the tech-savvy among us).

We are going to use Make.com (formerly Integromat) as our orchestrator.

Why Make and not Zapier? Because for this specific workflow, Make’s ability to filter arrays and handle JSON data is significantly cheaper and more visual. We need to parse calendar data, which is often messy. Zapier can do this, but you will pay a premium for the extra steps. If you are serious about automation in 2026, Make is the standard for complex logic.

Here is the flow we are building today:

1. Trigger: Google Calendar detects a new event with an external email address.
2. The Brain: We send that email domain and name to the Perplexity API (the hidden gem of modern research).
3. The Filter: We instruct the AI to ignore marketing fluff and look for friction or money (recent funding, layoffs, new product launches).
4. The Delivery: The summary arrives in a private Slack channel or updates a Notion database property just before the call starts.

The Stack (What You Need)

You can set this up on the free tiers of most of these tools, but for reliability, small paid plans are recommended.

1. Make.com Account: The glue holding it all together.
2. Google Workspace: The source of truth for your schedule.
3. Perplexity AI (API): We are using Perplexity instead of standard OpenAI because Perplexity is built for live web browsing and citation. Standard GPT-4 models can be lazy about browsing; Perplexity is aggressive about it.
4. Slack or Notion: Where you consume the data.

Step 1: The “Bouncer” (Google Calendar Trigger)

First, we need to ensure this automation does not fire for your internal stand-up meetings, your lunch breaks, or your dentist appointment. If you skip this, you will burn through credits researching your own coworkers.

1. Create a new Scenario in Make.
2. Add the Google Calendar module.
3. Select “Watch Events.”
4. Set the limit to “Future Events.”

The Configuration Setting That Matters This is where most people fail. They connect the calendar and immediately let it run wild. You must add a Filter immediately after the Calendar module. Click the dotted line connecting the Calendar to the next step to set this up.

• Label: External Guests Only
• Condition: Attendees: Email [Does not contain] @yourcompany.com

Alternatively, if you only want this to run for sales calls, you can filter by Event Title [Contains] “Discovery” or “Demo.” This ensures the automation only wakes up when money is on the table.

Step 2: The Data Prep (Isolating the Target)

Calendar data is notoriously messy. An “Attendee” object usually contains a name, an email, and a status (accepted/declined). We need to isolate the domain of the company and the name of the most important person.

If you skip this step, the AI gets confused by the clutter.

1. Add a Text Parser module (or use Make’s built-in text functions).
2. We want to extract the domain from the email address.
3. Use the split function on the email address, splitting by the @ symbol, and take the second item.

Pro Tip: If the email is generic (gmail.com, yahoo.com), you need a router here to stop the automation. There is no value in researching “gmail.com.” Set a filter that says: Domain [Does not equal] gmail.com.

This small step prevents your “Private Investigator” from returning a report on Google Alphabet Inc. whenever you meet someone with a personal email address.

Step 3: The “Private Investigator” (Perplexity API)

Here is where we deviate from the standard “ask ChatGPT” advice.

ChatGPT is incredible for synthesis, but for live reconnaissance, Perplexity is superior because of its citation index and real-time crawling capabilities. It doesn’t just hallucinate; it reads the news from this morning.

1. Add an HTTP Module in Make.
2. Set the method to POST.
3. URL: https://api.perplexity.ai/chat/completions (Check their current documentation for the exact 2026 endpoint).
4. Headers:

• Authorization: Bearer YOUR_API_KEY
• Content-Type: application/json

Now, we configure the payload. This is the “Secret Sauce.”

If you just ask “Who is John Doe?”, you will get a bio. We want leverage. We want to know what problems they are facing so we can solve them.

The “Sherlock” Prompt Template

Copy and paste this into the content section of your JSON payload. Note how specific the instructions are regarding negative information (hiring freezes, bad PR). That is often more useful than positive news.

{
 "model": "llama-3-sonar-large-online",
 "messages": [
 {
 "role": "system",
 "content": "You are an expert business analyst and private investigator. You value brevity, financial data, and recent news over generic marketing fluff."
 },
 {
 "role": "user",
 "content": "Analyze the following lead based on their meeting invite data.\n\nTarget Name: {{Name from Calendar}}\nTarget Company Domain: {{Domain from Step 2}}\n\nTask:\nResearch the company and the person. Provide a 'Dossier' strictly following this structure:\n\n1. THE HOOK: One specific piece of news from the last 90 days (funding, product launch, layoffs, or PR crisis).\n2. THE HUMAN ELEMENT: Check LinkedIn/public data. What is the person's specific role? Do they post about specific technical problems? (If no data, say 'No public profile found').\n3. THE RED FLAG: Is the company hiring aggressively (growth) or freezing hiring (trouble)?\n4. THE ICEBREAKER: One non-business fact if available (alma mater, volunteer work), otherwise a mutual interest based on industry trends.\n\nConstraints: Keep it under 200 words. Use bullet points. Do not use introduction or conclusion text."
 }
 ]
}

Why this prompt works: It forces the AI to look for changes (news from the last 90 days). Sales and partnerships happen when things change. If a company just raised Series B funding, you pitch growth. If they just laid off 10% of staff, you pitch efficiency.

Step 4: The Delivery Mechanism

Where do you actually look during your workday?

If you are a Project Manager living in Notion, put the data there. If you are in Sales and live in Slack, send it there. For this tutorial, we will use Slack because the notification factor is helpful right before a call.

1. Add the Slack module.
2. Select “Create a Message.”
3. Choose a private channel. I call mine #meeting-intel or #dossier-bot.
4. In the text field, map the output from the Perplexity module.

The “Hidden Gem” Configuration: Slack blocks allow for formatting. Don’t just dump the text. Make it readable.

Add a “Button” in the Slack message block that links directly to the LinkedIn Search URL for that person.

• Button Label: “Verify on LinkedIn”
• URL: https://www.linkedin.com/search/results/all/?keywords={{Name from Calendar}}

This saves you the inevitable step of alt-tabbing to check their face before the camera turns on. It reduces friction.

Step 5: Handling the “Ghost” (Error Handling)

Sometimes, the AI finds nothing. Maybe the person has zero digital footprint, or the company is in stealth mode.

If you don’t handle this, your automation might break or send you a hallucination.

In Make, add an Error Handler (Right-click the HTTP module -> “Add error handler”). If the API fails or returns a 404 on the search, have it send a fallback message to Slack: “Target is off the grid. No digital footprint found. Proceed with caution.”

This is actually valuable intel. It tells you that you are dealing with a privacy-conscious individual or a very early-stage startup. Knowing what you don’t know is just as important as the facts.

Why This System Beats “Hustling”

I have seen sales teams hire juniors just to do this work (they call them SDRs). I have seen Project Managers spend their Sunday nights prepping for Monday reviews. We treat this burnout as a badge of honor, but it is actually a failure of systems.

When you implement this, two shifts happen in your workflow:

1. You Stop Panicking: You know that 15 minutes before the call, the information will come to you. You don’t have to go hunting for it.
2. You Build Authority: When you open a call by saying, “I saw you guys just launched that beta feature last week, how is the server load handling it?” rather than “So, what do you guys do?”, the dynamic changes instantly. You aren’t a vendor anymore. You are a peer.

The Cost vs. Value Equation

Let’s look at the math. Make.com: Free tier (or $9/mo). *Perplexity API: Roughly $0.01 per search. Your Hourly Rate: Likely $50, $100, or more.

If this automation saves you 15 minutes of frantic Googling per meeting, and you have 10 external meetings a week, that is 2.5 hours saved. That is pure ROI.

The Next Level: Closing the Loop

You now have a system that ensures you know everything about them before the meeting starts. But there is a second half to this equation that most professionals ignore.

What do they find when they research you?

Just as your “Dossier Bot” is scouring the web for their information, they are likely looking you up on LinkedIn or Google.

If your automation gives you a competitive edge in information, your personal branding gives you a competitive edge in perception.

In 2026, text-based bios are becoming invisible. We scroll past them. The professionals winning the biggest contracts and the best roles are the ones who control their narrative visually.

While your Make.com automation handles the input (learning about them), PortfolioVideo handles the output (teaching them about you). It allows you to transform your static resume or pitch deck into a dynamic, AI-narrated video presentation.

Imagine this workflow: 1. Input: Your Dossier Bot tells you the client is worried about “Scaling Efficiency.” 2. Output: You reply to the meeting invite with a 60-second, AI-generated video intro where your digital avatar specifically mentions your experience with scaling.

You didn’t just research them; you proved you are the solution before the call started.

The “Next 15 Minutes” Challenge

Do not just bookmark this article. Open Make.com right now. Connect your Google Calendar. Add the filter. Even if you don’t build the full AI integration today, just getting the data flowing is the first step.

Build the exoskeleton. Stop doing the manual labor.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI