AI Projects · Lesson L05
The Online Business Stack for 2026: A Small Operator's Field Guide
A small online business in 2026 is mostly automated scripts. Here is the stack — landing page, product, email, monetization, analytics — and which pieces to automate first.
Last tested and updated: June 2026
A one-person online business in 2026 needs five things: landing, product, email, payment, dashboard. By week three, four of those five run on their own. Scripts fire on sign-up, post a weekly digest, and ping you on traffic drops.
What does the whole stack look like, and which piece do you automate first? See Hermes L02, L01, and L04 for the prerequisites.
The hook
A friend runs a small business that’s been going for about four months. He works on it roughly six hours a week and makes more than his last salary.
What does he actually do in those six hours?
- Monday: review the weekly digest his Hermes agent sends. Approve three draft posts the agent queued.
- Wednesday: ship one improvement to the directory, prompted by a user email.
- Friday: record a 10-minute Loom for the email list. The transcript gets transcribed and summarised by an automated skill before the week is out.
- Ad-hoc: reply to a few customer emails. Anything that looks like a recurring question becomes a Skill, not a personal reply.
The landing page, the listings scraper, the welcome sequence — all of it runs on cron jobs and skills. The owner’s job is judgement calls, not button-pushing.
Name one routine task you do more than once a week. You can replace it with a script by this time next month. That’s the whole game.
The mental model
A real online business in 2026 has five layers. Each depends on the one below it. Each has a different automation profile.
Layer 1 — Landing. The page that captures an email. Static site, Carrd, or one Next.js route. Cost $0–$10/mo. Automation potential: low — write the copy once, hook the form to your list.
Layer 2 — Product. What you deliver. The directory, the newsletter, the app. Automation potential: medium — Claude Code shines here. The work lives in a repo and needs test verification (see L01).
Layer 3 — Email. The list, welcome sequence, weekly digest. The highest-leverage layer. Every other layer funnels back into it. Automation potential: very high — a welcome is a trigger; a digest is a cron.
Layer 4 — Monetization. Stripe links, Substack, paid placements. Automation potential: medium — webhooks fire scripts (“send receipt,” “add to premium list”). You never manually invoice.
Layer 5 — Analytics. Plausible, Umami, or a cron-driven report. Automation potential: very high. The dashboard updates itself. The only automation worth building is the alert. A Hermes cron that scans the dashboard daily and pings Discord if traffic drops 30% week-over-week has saved this business twice.
Where the time goes
| Activity | Hours/week | Automated? |
|---|---|---|
| Writing / recording | 3 | No |
| Reviewing agent output | 1.5 | Partially |
| Customer email replies | 1 | Partially |
| Strategic decisions | 0.5 | No |
| Everything else | 0 | Yes |
The “everything else” row is the win. If your week has any row that isn’t writing, deciding, or reviewing, you’re doing something a script should be doing.
Pick your tool
Here’s the stack I’d actually build this week, in order.
Week 1: The minimum. Set up Layers 1, 3, and 4 with Hostinger Horizons (or Carrd for free), Buttondown or ConvertKit for email, and Stripe Payment Links for monetization. Cost: $0–$15/mo. No automation yet — just the bones.
Week 2: Add a coding harness. Use Claude Code to build Layer 2 — the directory MVP from L04, newsletter archive, or SaaS dashboard. Hermes is the wrong tool here. The work lives in a repo and needs test verification.
Week 3: Add an agent harness. Install Hermes and automate Layers 3 and 5: tag new sign-ups by source. Add a Monday-morning cron (see Hermes L08) that posts a digest to Discord. Add a daily analytics scan that pings only on anomalies.
Week 4: Wire it together. Stripe webhooks fire Skills. New customers get personalised welcome emails. The directory auto-updates with listings scraped from a niche subreddit. You stop opening the dashboard unless the agent pings you.
What you should NOT build first
Most beginners jump to AVOID or SKIP quadrants. Examples: a chatbot for support (high effort, low payoff), daily LinkedIn posts (audience isn’t yours), a custom CRM (buy one — the directory is your CRM). The bottom-right quadrant is where your week-one budget goes: welcome sequence, payment receipt, weekly digest. Cheap to build, runs forever.
| Layer | Tool | Cost | Time to automate |
|---|---|---|---|
| Landing | Hostinger Horizons | $5/mo | 2 hours |
| Product | Claude Code + your stack | BYOK model | 1–2 weeks |
| Buttondown / ConvertKit | Free–$15/mo | 1 day | |
| Monetization | Stripe Payment Links | 2.9% + 30¢ | 30 min |
| Analytics | Plausible + Hermes cron | $0–$9/mo | 2 hours |
| Orchestration | Hermes on a small VPS | $3–$5/mo | Ongoing |
Total: $15–$35/mo + model API costs. Models and your time are the real cost.
Try it
The exercise
Pick ONE thing from the DO FIRST quadrant and ship it by Sunday. Not all four. One.
The easiest first automation is a welcome email sequence. Sign up for Buttondown. Write three emails (intro, best work, a question). Set a trigger. Email 1 fires immediately. Email 2 fires after 48 hours. Email 3 fires after 7 days.
Bonus: install Hermes and write a Skill that tags each subscriber by source. A Skill is a saved prompt that runs the same way every time.
If you can ship this by Sunday — even a crude version — you have the foundation for everything else. Every other automation follows the same shape: a trigger, a sequence of steps, and a log.
Check your understanding
See quiz.json — six questions extracted for Quiz.tsx.
What’s next
You’ve completed the AI Projects module. To go deeper: