At some point last year I had three active projects running simultaneously, a day job, and a kid who wanted to validate a Lego business idea on a Saturday afternoon.
Something had to give — and it wasn't going to be the kid.
So I got serious about no-code automation for small business. Not because I read a blog post about it. Because I was drowning, and building my way out was the only thing that made sense.
Here's the stack I actually use. What I pay for, what I don't, and what I'd hand to someone starting from zero today.
Why I Stopped Manually Doing Anything That Could Be Automated
The turning point was Shy Car. I was manually pulling car listings from multiple sources, formatting them, and pushing them to a website. Every. Single. Day.
It took me about a week of doing that before I snapped and said "there has to be a webhook for this." (There was. More on that in how I automated car listings with AI.)
The thing about manual work is it expands to fill your time. You do it because it feels productive, but it's not — it's just busy. The real work is building the thing that does the work.
I now have a rule: if I do something more than twice and it takes more than 10 minutes, I find a way to automate it before the third time.
That one rule changed how I build.
The Core Stack (What I Actually Pay For)
I'll keep this honest. Here's the stack I'm paying for monthly, in order of "would stop everything if it disappeared":
Make.com (formerly Integromat)
This is the engine. Make is where I build most of my no-code automation for small business use — connecting apps, triggering actions, processing data. It's more powerful than Zapier for complex flows, and at my usage level, meaningfully cheaper.
For Shy Car, Make pulls car listing data, passes it through a format step, and pushes it to the site automatically. For 1 Yr Net, Make runs the data aggregation that makes the dashboard useful. It's behind almost everything I do.
What I'd say if you asked me: start here. The learning curve is real but short, and it pays you back fast.
Ghost (self-hosted)
My blog platform for everything: Age of Robots, the Harbor Soils blog I run for a landscaping client. Ghost does one thing well — it gets content on the internet without fighting me. No plugin nightmares, no bloat. I publish, it shows up.
Zapier (limited, legacy)
I'll be honest: I still have a few old Zapier automations running that I haven't migrated yet. It works. But if I were starting today I'd go straight to Make.
OpenClaw / AI tooling
This is less "one tool" and more "a layer." I use AI to draft content, QA structured data, and assist with research. The key word is assist. I'm not having AI write my posts wholesale — I'm using it to accelerate steps I'd do anyway.
AI vs. No-Code: What's the Difference and When Does It Matter?
People mix these up constantly, so let me draw a clear line.
No-code automation is logic: "when this happens, do that." It doesn't think, it executes. Make.com moving data from a Google Sheet to a website is no-code. It's conditional logic and integrations, not intelligence.
AI is inference: "given this input, generate an output based on patterns." It makes decisions — fuzzy, probabilistic decisions. Writing a product description, extracting intent from a support email, turning a VIN number into a car listing summary. That's AI.
The real power is when they work together. For Shy Car: the webhook triggers (no-code), pulls the raw data, passes it to a language model for formatting and summary (AI), then pushes the clean result to the site (no-code again). Neither one alone would do the job.
How Each Tool Fits Into a Real Project (Not a Hypothetical)
Shy Car — The core automation is Make + webhooks + an AI formatting step. Car listing data comes in messy. It goes out clean and publishable. I check it maybe twice a week now instead of daily.
1 Yr Net — This is my habit and goal tracking project. The automation here is simpler: Make pulls data from a few sources and aggregates it into a format I can actually read.
Simply Lawn — A directory site I built to learn Brilliant Directories and no-code directory tooling. There are Make scenarios running behind it for lead routing and listing notifications.
Each project taught me something different. Shy Car taught me webhooks. 1 Yr Net taught me data normalization. Simply Lawn taught me that "no-code" still requires you to understand the business logic — the platform just removes the syntax barrier.
What I'd Add Next (and What I'd Cut If I Had to)
What I'd add: A better monitoring layer. Right now I'm checking automations manually. I want something that pings me when a Make scenario fails silently. This is a real gap in my stack.
What I'd cut if budget got tight: Zapier, immediately. Ghost, reluctantly. Make, never. That's the engine. Cutting it would mean rebuilding manually.
If You're Starting From Zero
Here's the honest answer: don't start with a stack. Start with a problem.
What's one thing you do manually every week that you hate? Map it. Then find the single tool that removes that step. Don't build the full automation architecture first — you'll build the wrong one.
For most solopreneurs, that first tool is either Make.com (if your problem is connecting apps) or a scratchpad AI (if your problem is writing/research volume). Start with the one that matches your actual bottleneck.
Once that's working, add the next layer. My stack didn't look like this six months ago. It won't look like this six months from now. That's the point.
More from the Age of Robots build log: The Indie Hacker's Honest Guide to AI Tools in 2026 — what I'm actually paying for and what's overhyped. And if webhooks confuse you: Webhook Automation for Non-Developers.
