I Automate Everything That Doesn't Need Human Judgment -- Here's What That Looks Like

I run three lines of work: domain investing, freelance web development, and an e-commerce business. None of these are “passive.” But I’ve automated enough of them that I spend my time on decisions, not execution.

Here’s the philosophy: if a task takes less than 30 seconds of human thinking, it should be automated. Not because I’m lazy, but because every minute spent on predictable work is a minute I’m not selling, creating, or improving.

What I Automate and How

Domain appraisals. I used to check each domain manually: DNS, Wayback, NameBio. 5 minutes per domain. I processed 20 domains a day. That’s 100 minutes. Now my daily pipeline script checks DNS, scores domains, looks up comparable sales, and ranks them. I review the output in 5 minutes.

Cold email sending. Writing and sending 50 personalized emails used to take 2 hours. Now I have templates per domain with variable fields, a contact dedup system, and a rotary sender that rotates between email providers. I spend 10 minutes verifying the contact list, the script sends over an hour.

Content posting. I used to copy-paste between Twitter, Bluesky, and Dev.to. Now OmniPost generates content, posts on schedule, and cross-platforms automatically. I review the queue once a day.

Delivery price calculation. Every time we quoted a delivery price, someone had to look it up. Now it’s calculated instantly from a lookup table. 30 seconds per order saved, 30+ orders a day, that’s 15 minutes daily.

The Automation Decision Tree

Before I automate anything, I ask three questions:

1. Does this task require human judgment? A cold email template doesn’t. Reading a reply does. Appraising a domain doesn’t (I’ve encoded the criteria). Deciding whether to register it does.

2. How much time does it consume weekly? Less than 30 minutes? Don’t automate. More than 2 hours? Automate immediately. In between? Calculate the setup time and build the automations with the highest ROI first.

3. Will the automation break often? If the data source changes weekly, the automation will need constant maintenance. That’s not automation, that’s a second job. Only automate things that stay stable.

What NOT to Automate

Some things I deliberately keep manual:

• Client conversations. Automated outreach is fine. Once a human replies, the conversation is human-only.
• Portfolio decisions. Which domain to buy, which project to take, which wilaya to expand to. These are judgment calls that should consider factors no script can capture.
• Strategy adjustments. What’s working this month? What changed? Automation handles execution. Strategy is a human function.
• Content ideas. I generate content concepts manually. The writing and posting can be automated, but the “what should I write about” question is human.

The Tools

Here’s my actual automation stack:

• Cron jobs (Hermes cron): Schedules all recurring tasks. Daily pipeline at 8 AM, content posting hourly, database backup at 3 AM.
• Python scripts: Heavy lifting. Domain appraisal, email sending, contact management, data analysis.
• MCP servers: API access for agents. Domain sales DB, email infrastructure, GitHub.
• Google Sheets API: Connects the e-commerce team’s operational data to my automation layer.
• GitHub Actions: Deploys the site, runs scheduled exports, validates data.

All free-tier. All glue code I wrote in an afternoon each.

How You Can Do It Too

Start with one 30-minute task you hate. Not the most important task, the one you dread doing. The one you put off.

For me, it was checking domain availability. I hated opening 5 tabs per domain. I automated that first. The momentum from that one automation made me look for the next one.

Write the script, don’t buy the tool. Most “automation platforms” (Zapier, Make) cost money and have limits. A Python script costs $0, runs on your machine, and does exactly what you need.

Set it and monitor it. Every automation breaks eventually. Set up monitoring (I use cron notification) so you know when it fails.

Related: Automation 200 Lines

##s

Q: How much time do automations actually save me?

A: About 10-15 hours per week across all three businesses. But the bigger benefit is consistency – the automation never forgets, never procrastinates, never makes typos.

Q: What’s the most valuable automation I’ve built?

A: The domain daily pipeline. It processes 20,000 domains, filters to the top 5, and presents them to me in 45 seconds. Finding one good domain pays for a year of automation maintenance.

Q: Do I need to know how to code?

A: Yes, for custom automations. But AI coding agents can build them with guidance. Describe what you want, review the code, test it.

Q: What’s the biggest automation mistake?

A: Automating something that’s broken. If the manual process is wrong, automation makes it faster to produce wrong results. Fix the process first, then automate.

Related: How I Use AI Agents as My Coding Partner – the agent layer above automation. Building OmniPost – a specific automation in detail.