Automate Sales Follow-Up with Screenpipe: The No-Code Lead Tracking System

What You Will Get: Automate Follow Up Without a Developer

You just finished a sales call. The prospect said "let's discuss pricing." You planned to send a calendar link. Then the email notifications piled up. Three days later you realize you never followed up. They booked a demo with your competitor instead.

That scenario happens every day. The fix is not to try harder. The fix is to automate sales follow-up automation at the moment the trigger happens. Screenpipe, an open-source screen and audio recorder, captures every on-screen text and window title on your laptop. It sends that data via webhook to Zapier or Make. Within seconds you can create a deal in HubSpot and send a personalized email with your Calendly link. No developer. No monthly subscription for Gong or Fireflies.

This system turns every demo, webinar, or casual conversation into a logged lead with a time-stamped follow-up. The result is faster response, more booked calls, and hours reclaimed from manual data entry. InsideSales data shows that following up within 5 minutes makes you 100x more likely to connect. Screenpipe makes that speed possible without a dedicated ops person.

What You Need: Tools and Setup Checklist

Before we walk through the steps, here is the exact checklist of tools and permissions you need. Every item is free or has a generous free tier.

• Screenpipe (free, open-source). Install it on the laptop where you run sales calls and demos. It processes all data locally. No vendor lock-in. The GitHub repo uses an MIT license.
• Zapier or Make (free tier works for low volume). This is your no-code glue. It receives the webhook from Screenpipe and triggers actions in your CRM and email.
• HubSpot CRM (free tier supports webhook-based deal creation). Any CRM with a webhook API works, but HubSpot is the easiest for this tutorial.
• System permissions for screen recording and audio capture. You must disclose to leads that you are recording. This is not optional. Check your local consent laws.

That is it. You do not need a developer. You do not need to touch a line of code beyond one copy-paste command. Spend 30 minutes on this Screenpipe setup guide and you save hours every week.

Step 1: Install and Configure Screenpipe with a Webhook Trigger

Open your terminal (macOS or Linux). Run this single command:

brew install screenpipe

If you do not have Homebrew, download the binary from the GitHub releases page. After installation, start Screenpipe with the webhook flag pointing to your Zapier catch hook URL. Replace the example URL with your own.

screenpipe --output webhook --webhook-url https://hooks.zapier.com/hooks/catch/your-key/

This tells Screenpipe to send every detected on-screen text event as a JSON payload to that Zapier URL. The capture runs at about 2 frames per second. That is enough to catch window titles, chat messages, pricing mentions, and "next steps" phrases.

Now configure which windows to capture and which to exclude. You only want the Zoom window, your browser with the CRM, and maybe your email. Exclude private apps like messaging or password managers. Add this to your config (create a screenpipe.config.yaml file):

include: ["Zoom", "Chrome", "Firefox"]
exclude: ["Slack", "Messages", "1Password"]
storage_rotation_days: 7

This Screenpipe webhook trigger will run silently in the background. It consumes about 1 GB of disk per hour as plain text logs. Set rotation to 7 days so you do not run out of space. I have seen founders forget this and fill a 256 GB drive in two days. Set a weekly clean script or simply delete the logs folder manually every Friday.

Understanding the Data

The JSON payload includes the window title, on-screen text (via Tesseract OCR), and a timestamp. OCR accuracy is 85 to 95 percent depending on font and contrast. Names and email addresses are the trickiest. Do not rely on 100 percent capture. Always allow a manual override field for critical data.

If you handle sensitive data like patient records or credit card numbers, mask those windows in the exclude list. Recording screen content without consent is illegal in many jurisdictions. Disclose clearly at the start of every call: "I record my screen for note-taking and follow-up. The data stays on my machine. Is that okay?"

Step 2: Build a No Code Automation in Zapier

Now the fun part. You will build a Zapier sales automation that turns a raw text capture into a logged deal and an email blast.

1. Create a new Zap. Choose Webhooks by Zapier as the trigger and select "Catch Hook." Copy the webhook URL. Paste it into your Screenpipe config (as shown above).
2. Add a Filter by Zapier step. You only want events that contain your trigger keywords. Use phrases like "pricing," "next steps," "send proposal," or "quote." Make the filter case-insensitive. This prevents false positives from unrelated chatter.
3. Add an action: HubSpot > Create Deal. Map the captured text to the deal name. Set the deal stage to "Follow Up Needed." Include the timestamp and the prospect name if you can extract it. If OCR missed the name, leave a note that says "manual name entry required."
4. Add a second action: Gmail > Send Email. Use a template that says "Thanks for the chat. Here is my calendar link to continue our discussion." Insert your Calendly or Chili Piper link. Personalize with the captured company name if available.
5. Insert a Delay by Zapier step before the email. Set it to 15 minutes. You do not want the prospect to receive an email while you are still talking to them. That feels robotic and desperate.

Test the Zap by running a short demo with a colleague. Say "let's discuss pricing." Check Zapier history. You should see a new deal created and a queued email. The whole loop takes about 10 seconds. Speed to lead drops from hours to minutes.

Realistic numbers for a solo SDR: 20 demos per week, 50 percent trigger rate. That is 10 automated follow-ups per week. Each manual follow-up used to take 10 minutes. You save nearly two hours weekly. Over a quarter that is a full week of selling time reclaimed.

Pitfalls to Avoid (and How to Fix Them)

This system works, but it is not magic. Here are the common Screenpipe pitfalls that operators hit and how to dodge them.

OCR Accuracy Is Not 100 Percent

Screenpipe uses Tesseract OCR. On a clear screen with Arial font at 14px it hits 95 percent. Throw in a custom font over a dark background and accuracy drops to 80 percent. Never assume the captured email address is correct. Fix: Add a manual review step in Zapier. Send a Slack notification with the captured text and a "confirm" button before creating the deal. Or simply accept that 1 in 10 deals will need a quick edit.

Storage Fills Fast

1 GB per hour of screen logs is real. If you run Screenpipe all day, that is 8 GB daily. A 256 GB drive fills in a month. Fix: Set storage_rotation_days: 3 in the config. Or compress logs weekly with a cron job. Better yet, only run Screenpipe during calls, not all day. Use a hotkey to start/stop it.

False Positive Keyword Triggers

The word "pricing" appears in many contexts. "The pricing page is live" or "I will check pricing later" are not buying signals. Fix: Use a two-word proximity filter in Zapier. Require "pricing" within 50 characters of "discuss" or "send." Even better, pipe the raw text through a local LLM via Ollama. Ask the LLM: "Does this segment contain a negotiation or proposal request?" That cuts false triggers by 80 percent. The latency is about 10 seconds per check, fine for non-urgent flow.

Legal Risk

Screen recording without consent is illegal in two-party consent states (California, Florida, Illinois) and many EU countries. Audio recording is even more regulated. Fix: Disclose at the start of every interaction. Use a verbal script. Include a checkbox on your booking form. If the prospect objects, stop Screenpipe for that call. The lost data is better than a lawsuit.

Where to Go Next: Scale and Optimize

This single-laptop setup proves the concept. When you want to move from solo to team, you need a different tool. Conversation intelligence automation at scale means centralizing data. Gong or Fireflies provide team-wide dashboards, automated coaching, and rep-level analytics. Screenpipe is the builder's choice for privacy and zero cost. But it does not give you visibility into your team's calls.

For the intermediate step, keep Screenpipe but add an LLM summary layer. Use Ollama (local) or the OpenAI API to generate a call summary and action items from the captured screen and audio. Screenpipe can output a full transcript. Pipe that to an LLM prompt: "Summarize this call in three bullet points and list one follow-up action." Store the summary in your CRM notes field. This makes your log instantly scannable.

A/B test your trigger keywords and follow-up delay. You might find that "send proposal" triggers more bookings than "pricing." Or that a 5-minute delay converts better than 15 minutes. Run the experiment for two weeks. Measure booked calls per trigger. Adjust accordingly.

If you want to see exactly where your current site and funnel leak leads, run a free AI audit. It takes minutes and shows you the gaps in your tracking and follow-up flow.

Building this yourself is a solid weekend project. But if you would rather skip the setup and get a done-for-you system that includes the tracking, the automations, and the follow-up sequence, we offer a Growth Sprint. It is a fixed-scope two-week build: landing page, tracking setup, automated follow-up, no retainer. Learn more at our offer page. No pressure. The steps above will get you 80 percent of the way there on your own.

Cover photo by Pachon in Motion on Pexels.