I Built a Content Empire While Sleeping — Here’s the Secret

#automation #ai #python #tutorial

I Built a Content Empire While Sleeping — Here’s the Secret

I’ve always thought of myself as a “night‑owl creator.” After a full‑time dev job, I’d spend the evenings tweaking blog posts or brainstorming video ideas. The problem? By the time I finally hit “publish,” my brain was already on the next day’s sprint. I wanted more content out there, but I didn’t have the energy (or the time) to film, edit, and upload every single day.

That’s when I stumbled on the phrase AI video automation while scrolling through a Reddit thread. The idea of a bot that could write scripts, find images, add a voiceover, and spit out a polished short video felt like a sci‑fi plot twist. I was skeptical, but also desperate enough to try. This post is the journal of that experiment—wins, setbacks, and the exact workflow that now runs on autopilot while I’m sleeping.

Week 1: Sketching the Dream (and Realizing I Needed an n8n workflow)

My first step was to map out the whole process on paper:

Idea generation – a list of topics I care about (tech trends, AI tools, dev‑life hacks).
Script writing – using an LLM to turn bullet points into a 60‑second script.
Image & footage sourcing – searching royalty‑free libraries automatically.
Voiceover – text‑to‑speech that sounded natural enough for a quick watch.
Video assembly – stitching everything together, adding on‑screen text.
Auto‑posting – push to YouTube Shorts, TikTok, and Instagram Reels.

I’d heard of n8n, an open‑source workflow automation tool, but I had never built a full‑fledged content automation pipeline with it. The learning curve was steeper than I expected. The first 4 hours were spent just installing Node.js, Docker, and trying to get n8n to run locally. I ended up with a “Cannot connect to database” error that lasted a whole afternoon.

Mini‑setback #1: My dev instincts told me to keep digging into the docs, but the clock was ticking (it was already 11 pm). I decided to pull an overnight break, come back fresh, and treat the workflow like a small side‑project rather than a production system.

Week 2‑3: Building the First Prototype (AI Shorts in Action)

Armed with fresh coffee and a clearer mind, I started connecting nodes:

Trigger: A Google Sheet where I’d drop one‑line ideas each morning.
LLM Node: OpenAI’s API to generate a 150‑word script.
Image Search Node: Pexels API to fetch 3–4 relevant images.
Voiceover Node: ElevenLabs for realistic TTS.
FFmpeg Node: Concatenate voice, images, and subtitles into a 1080p video.
Publish Node: The YouTube API, TikTok’s upload endpoint, and Instagram’s Graph API.

I called the whole thing AI Shorts because each video was meant for the short‑form platforms. After a few test runs, the output video looked passable, but the subtitles were mis‑timed, and the voice sounded a little robotic. I spent a day fine‑tuning the prompt I sent to the LLM (“Write a concise, conversational script for a 60‑second video about X”) and added a small delay node before the voice clip to sync with the images.

Mini‑setback #2: The first time I let the workflow run unsupervised, it posted a video with a broken thumbnail to TikTok. The platform flagged it, and my account got a temporary warning. I learned to add a “validation” step that checks the thumbnail URL before publishing.

Despite the hiccups, the prototype was functional. I scheduled the workflow to run at 2 am UTC, so my inbox would fill with fresh Shorts while I was still dreaming.

Week 4‑5: Scaling Up (Passive Income AI Starts to Feel Real)

With the core pipeline solid, I turned my attention to automated video production at scale. I added a few niceties:

Keyword Research Node: Using Ahrefs API to pull high‑search‑volume keywords related to my niche.
Title Generator Node: LLM prompts that created click‑bait‑friendly titles (e.g., “How AI Created a 1‑Minute Video in 10 Seconds”).
Analytics Node: Google Analytics and YouTube Data API to log view counts, watch time, and CPM.

Within 30 days, I had produced 84 AI Shorts, averaging 350 views each. The aggregate watch time bumped me into the YouTube Shorts Partner Program, and I started seeing roughly $15–$25 per week in ad revenue—tiny, but it felt like real passive income AI. More impressively, the workflow kept churning out content with zero manual effort after the initial idea drop.

I also tweaked the voiceover provider to a slightly more expensive tier, which cut the robotic tone in half. The upgrade cost $5/month but increased average watch time by 12%, a trade‑off I was happy to make.

Week 6‑8: Polishing the Experience (From Prototype to Product)

By the eighth week, the workflow felt polished enough that I wanted to share it with the community. I stripped out my personal API keys, added a few UI prompts (like a dropdown to choose the publishing platform), and bundled everything into a single n8n workflow file.

I posted the workflow on my GitHub and wrote a short guide on how to set it up. The response was surprisingly enthusiastic—people asked for a “one‑click” version that handled script generation, voice, and video without needing to juggle multiple API keys.

That’s when I realized I could turn this into a small product. I packaged the workflow, added a couple of pre‑configured API credentials (via a secure secrets manager), and set a one‑time price of $20. The final offering is called AI Shorts Factory and includes:

Everything you need for end‑to‑end AI video automation (script, images, voice, video).
Auto‑posting to YouTube, TikTok, and Instagram.
A 30‑day support window for any setup issues.

Since launching, I’ve sold 25 copies in the first week. Not a fortune, but it validates that the idea solves a real pain point for creators who want content automation without spending weeks learning n8n.

The Takeaway: You Can Build a Tiny Empire While You Sleep

If you’re a developer with a side hustle, the biggest barrier to scaling content isn’t a lack of ideas; it’s the manual grind of turning those ideas into videos. An n8n workflow lets you encode the entire production pipeline into a repeatable, version‑controlled asset. Combine that with modern LLMs and high‑quality TTS, and you get a self‑sustaining engine that spits out short videos while you catch some Z’s.

My advice? Start small. Pick a niche, map out the steps, and build a proof‑of‑concept before you try to automate everything at once. Expect a couple of hiccups (broken thumbnails, timing issues) and treat them as part of the learning curve. Once you have a stable loop, you’ll be amazed at how quickly the numbers add up—both in views and in that sweet passive income AI stream.

The tool I’m using is called AI Shorts Factory (https://8622430312019.gumroad.com/l/gujqfy) — it’s an n8n workflow that costs $20 one-time and handles everything: AI script generation, image search, voiceover, video production, and auto‑posting to YouTube, TikTok, and Instagram.