I Built a Complete Marketing Automation Stack with n8n — 15 Workflows in Production

DNYANTRA — Thu, 26 Feb 2026 04:16:01 +0000

Six months ago, I was spending 3+ hours every day on marketing tasks for my SaaS product: hunting leads on Reddit, copy-pasting emails, writing blog posts, scheduling social content. All manual. All repetitive. All soul-crushing.

So I built an n8n automation system to do it for me. Now I have 15 workflows running in production, handling everything from lead generation to content publishing — with zero manual input.

Here's what I built and how it works.

The Architecture

Everything flows through a Google Sheets data hub — a 37-column master leads database that acts as the source of truth for the entire system. Every workflow reads from or writes to it.

The system has four pillars:

1. Lead Generation (5 Sources)

Five separate workflows scrape different platforms for potential customers:

GitHub scraper — finds developers actively building in your target niche by searching repos and issues
Reddit scraper — monitors specific subreddits for posts where people describe their pain points
Hacker News scraper — catches "Who's hiring" threads and Ask HN posts
YouTube scraper — finds creators in adjacent spaces who might need your product
ProductHunt scraper — identifies newly launched products for partnership/integration opportunities

Each scraper extracts: name, email (via Hunter.io), company, GitHub/social profile, and the exact post/content that triggered the lead.

2. AI Qualification

Every lead goes through a GPT-4 scoring workflow before entering the pipeline:

Input: lead profile + extracted content
Output: score 1-10 + qualification notes

Only leads scoring 7 or above proceed. This filters out noise — no more emailing people who clearly aren't a fit. The qualification prompt evaluates: relevance to your ICP, signal strength (are they actively looking for a solution?), and company size fit.

3. Cold Email Outreach

The email workflow is the most complex piece:

Email verification via MillionVerifier before any send (protects sender reputation)
IP warmup — starts at 20 emails/day, scales to 200 over 3 weeks using a Postal self-hosted mail server
Personalization — GPT-4 writes each email referencing the specific content that triggered the lead
3-sequence follow-up — automated follow-up emails at day 3 and day 7, each with different angles
Reply detection — stops sequences when someone replies

All tracked in the Google Sheet: sent date, open status, reply status, sequence position.

4. Content Generation & Publishing

Two blog generators (SEO-focused and founder-voice) that:

Pull topics from a content calendar in Google Sheets
Generate 1500–2500 word posts with GPT-4
Auto-publish to Ghost or WordPress via API
Trigger a social media workflow that creates platform-specific content (Twitter thread, LinkedIn post, Instagram caption) and schedules them

What This Actually Looks Like Day-to-Day

My morning routine used to involve 2+ hours of manual work. Now I open my Google Sheet, review the overnight leads (usually 20–40 qualified ones), spot-check a few emails that went out, and move on.

The system runs on a VPS with n8n self-hosted. Total infrastructure cost: ~$20/month. The ROI was immediate.

Lessons From Building This

1. Deduplication is critical. Without it, you'll email the same person 3 times from different scrapers. I use a hash of email+domain to deduplicate before anything enters the pipeline.

2. Email warmup can't be skipped. I burned a domain early on by sending 200 emails on day one. Start slow. The warmup workflow gradually increases send volume while monitoring bounce rates.

3. AI qualification changes everything. The quality difference between "all scraped leads" and "AI-qualified leads scoring 7+" is dramatic. Your reply rates will be 3–5x higher because you're only reaching people who actually have the problem you solve.

4. Google Sheets as a data hub works surprisingly well. I considered a proper database, but Sheets gives you a visual interface for free, easy filtering, and it's accessible to non-technical teammates. For this volume (<10k leads), it's perfect.

5. n8n > Zapier/Make for this use case. Self-hosted means no per-task pricing. Complex branching logic and custom JavaScript nodes make it possible to build things that simply aren't possible on no-code platforms.

The Tech Stack

Component	Tool
Workflow orchestration	n8n (self-hosted)
AI/LLM	OpenAI GPT-4
Email infrastructure	Postal (self-hosted SMTP)
Email verification	MillionVerifier
Email finding	Hunter.io
Data hub	Google Sheets
Blog publishing	Ghost / WordPress API
Social posting	Twitter API v2, LinkedIn API
Lead sources	Reddit, GitHub, HN, YouTube, ProductHunt APIs

Building This for Others

After running this for my own SaaS for 6 months and iterating until it actually worked, I started packaging this as a service on Fiverr.

If you want the full system — or just specific pieces like the lead scraper, email sequence, or blog generator — I've opened a gig: Build complete n8n marketing automation for leads, email & content

Packages start at $150 for a single workflow, up to $700 for the complete stack.

Happy to answer questions about any specific workflow — the Reddit scraper and the email warmup logic get the most questions, so drop them below.

I built a macOS Teleprompter with Electron (Don't kill me 😅)

DNYANTRA — Thu, 01 Jan 2026 20:43:06 +0000

Hi Devs! 👋

I'm Mayur. It's 2:15 AM here in Pune, India. 🇮🇳
I have to log into my "real" software engineering job in about 6 hours.

But I just shipped my side project, and I wanted to share the technical journey before I pass out.

The App: CamPrompter 🎥

It's a transparent teleprompter overlay for macOS. It lets you read scripts/notes while looking directly at the webcam lens during Zoom/Meet calls.

The Tech Stack: Electron ⚡️

I know, I know. "Electron is heavy." "Why not Swift?"

Honest answer: I wanted to ship fast, and I know JS better than I know Swift.

The Technical Challenge 🛠

The hardest part wasn't the text scrolling. It was Hardware Control.

I wanted to let users lock their Webcam Auto-Focus and Exposure directly from the app.

I had to bridge Electron with native UVC (USB Video Class) drivers to send hex commands directly to the webcam hardware.

The "Click-Through" Problem

Another pain point with Electron was making the window transparent and "click-through" so users can still interact with the Zoom window behind the text.

I managed to solve it, but I'm still optimizing the performance.

The Ask 🙏

I'm currently sitting at $0 revenue and running on caffeine.

If you are a Mac user (and not an Electron hater), I'd love for you to try it out and roast my performance.

The Link: https://camprompter.dnyantra.com/

Now, I am going to sleep. 😴
Happy coding!

DEV Community: DNYANTRA