5 min read
Picture a Tuesday evening. A new lead fills in your website form, an AI step sorts it as hot or cold, a draft reply lands in your inbox, and a row appears in your spreadsheet. It works beautifully, right up to the day your automation tool tells you that you have run out of tasks in the middle of the month.
That surprise is common, because the three big automation tools now charge for AI in three very different ways. So instead of comparing feature lists, let’s follow that one small workflow through Zapier, Make and n8n and see what it costs a solo business.
The test workflow: 150 leads a month
Here is our example. Every new form lead triggers five steps:
- The form trigger fires.
- An AI step classifies the lead.
- An AI-free step adds it to a spreadsheet.
- Another step creates a draft email.
- A final step sends you a notification.
So that’s one AI step and three ordinary actions after the trigger, 150 times a month. It’s small on purpose, because this is the size of automation most one-person businesses start with.
Zapier: easiest to start, and AI is multiplied
Zapier’s pricing page lists a Free plan with 100 tasks a month and two-step Zaps only, and AI by Zapier isn’t included there. The Professional plan starts at $19.99 a month billed annually, or $29.99 billed monthly, for 750 tasks. Triggers don’t count as tasks, and neither do failed actions.
The catch is AI. Zapier says each AI step is charged according to its model tier. One detailed pricing guide reports that since June 15, 2026, a step uses 1x, 3x or 5x the task rate depending on the tier, with the 3x Advanced tier as the default. Confirm the rate in your own Zap editor, because this is the one number that changes your bill.
Here’s our workflow on those rates:
- Standard tier (1x): 4 tasks per lead, so 600 tasks. That fits the 750-task plan.
- Advanced tier (3x): 6 tasks per lead, so 900 tasks. That doesn’t fit.
- Premium tier (5x): 8 tasks per lead, so 1,200 tasks.
In other words, leaving the default model on could push you out of the entry plan, while classifying a lead is the kind of simple job a Standard model should handle fine. That one setting is worth checking today.
Make: more room, trickier math
Make offers a Free plan with 1,000 credits a month, and most actions cost 1 credit. Routers and error handlers cost none. Its paid plans are Core, Pro and Teams, and the pricing page we checked didn’t display fixed prices in text, so look at the current slider for your usage.
AI is charged in one of two ways, according to Make’s credits documentation:
- Make’s AI Provider, on all plans: no outside account needed. Credits are based on tokens (small chunks of text the AI reads and writes) plus operations. The rate varies by model, and the small tier allows 18,080 input tokens per credit.
- Your own OpenAI or Anthropic connection, on paid plans: you pay Make 1 credit per operation and pay the AI company separately.
On a short lead message, our workflow should use about 5 credits per lead, or roughly 750 to 900 credits a month, which fits the Free plan on paper. The honest weakness is predictability. Token-based charges move with the length of what you feed the AI, so a workflow reading long emails will cost more than one reading short forms.
n8n: flat counting, but more of the work is yours
n8n counts differently: an execution is a single run of your entire workflow, however many steps it has. Our 150 leads are simply 150 executions. Its Starter plan is listed at €20 a month billed annually for 2,500 executions, and Pro at €50 for 10,000. That’s a huge amount of headroom for this workflow, and adding steps doesn’t change the count.
There are two trade-offs. You’ll generally connect your own AI account, which is a separate small bill. And n8n has a steeper learning curve for non-technical owners. There’s also a free self-hosted Community Edition, meaning you run the software on your own server, but “free” then comes with updates, backups and security becoming your job. If your site is already a source of late-night patching, as our WordPress patch checklist shows, think hard before adding a server to your list.
Side by side
| Question | Zapier | Make | n8n |
|---|---|---|---|
| What you’re billed on | Tasks | Credits | Whole workflow runs |
| Free option | 100 tasks, no AI by Zapier | 1,000 credits | Self-hosted Community Edition |
| Our 150-lead workflow | 600 to 1,200 tasks | About 750 to 900 credits | 150 executions |
| How AI is charged | Multiplier by model tier | Tokens, or your own key | Usually your own AI account |
| Best for | Speed and simplicity | Visual builders on a budget | High volume, comfort with tech |
So which one should you pick?
Here’s how I’d decide:
- Choose Zapier if you want it working this afternoon and you have fewer than a few hundred runs a month. Just set your AI steps to the cheapest tier that does the job.
- Choose Make if you’re happy spending an hour or two learning a visual canvas and want much more usage for less money.
- Choose n8n if your volume is high or your workflows have many steps, and you’re comfortable with a more technical tool.
Whichever you choose, remember the bigger saving is usually in the design. Ask whether the AI step needs to run on every record, or only on the ones a simple filter can’t sort. In our example, a plain rule could catch obvious spam before AI ever sees it. That’s the same thinking behind our quick wins for making AI actually save time.
Your 30-minute cost check
Before you switch anything, do this:
- Open your busiest workflow and count the steps.
- Note which steps use AI and which model they use.
- Multiply by last month’s runs, using the rules above.
- Drop one AI step to a cheaper model, then compare the results on ten real examples.
If the results hold up, you’ve just cut your bill without touching your tool. Prices and rates change often, so recheck before you commit to an annual plan. What’s the one automation you’d build first if the math worked?
