n8n vs Zapier vs Make for AI agents: what each really costs a small business in 2026
The three tools bill for different things. n8n charges per workflow run (an execution), however many steps it has: Cloud Starter is €20/month (billed annually) for 2,500 runs, or self-host the Community Edition for no licence fee. Zapier charges per successful step (a task). Pro starts at $19.99/month (annual) for 750 tasks, and AI steps can use 3–5 tasks each. Make charges per module action (a credit), including each polling check. Core is $9/month (annual) for 10,000 credits. For AI agents that run many steps on every item, per-run pricing (n8n) is usually the most predictable. For a few simple app-to-app automations, Make is the cheapest and Zapier is the easiest.
How each one bills
| n8n | Zapier | Make | |
|---|---|---|---|
| Billing unit | Execution = one full workflow run, any number of steps | Task = one successful action step. Triggers and built-in tools (Filter, Formatter, Paths…) are free | Credit = one module action, including reads and trigger checks |
| Free option | Community Edition, self-hosted (you run the server). Cloud trial: 1,000 executions | Free: 100 tasks/month, two-step Zaps only | Free: 1,000 credits/month, 2 active scenarios, 15-minute minimum interval |
| Entry paid plan | Starter €20/mo billed annually, 2,500 executions | Pro $19.99/mo annual ($29.99 monthly), 750 tasks | Core $9/mo annual ($10.59 monthly), 10,000 credits |
| Next step up | Pro €50/mo annual, 10,000 executions | Pro 1,500 tasks $39/mo annual ($58.50 monthly); 2,000 tasks $49/mo annual | Pro $16/mo annual ($18.82 monthly), 10,000 credits, priority execution |
| AI steps | AI Agent node included; bring your own OpenAI/Anthropic key, or use n8n's hosted AI credits | AI by Zapier: 1 task per step (standard or your own model), 3 (advanced), 5 (premium), plus tool calls | Your own OpenAI connection is a normal module; Make's own AI provider modules can use more credits |
| What happens at the limit | Upgrade, or buy extra executions on Pro-2 | Pay-per-task overage (1.25× base rate annual, 2.5× monthly) or Zaps pause | Scenarios stop until you add credits or upgrade |
Why the unit matters for AI agents: an AI agent rarely does one thing. A support agent reads the email, looks up the knowledge base, classifies, drafts and logs. That is 4–6 steps per email. On Zapier and Make every step is billed. On n8n the whole run counts once.
Worked example: an AI support inbox, 200 emails a month
Workflow: check the inbox → AI classifies and drafts a reply from your knowledge base → create a Gmail draft → log a row in a sheet. Assumptions (ours, not vendor figures): one AI call per email, a 30-day month, and nothing else running on the account.
| Checks every 15 min | Checks every hour | Plan that fits | |
|---|---|---|---|
| n8n Cloud | 2,880 executions (one per check, all new emails handled in that run) | 720 executions | Hourly fits Starter (€20). Every 15 min goes over Starter's 2,500, so Pro (€50), or self-host |
| Zapier | Trigger checks are free. About 3 tasks per email (AI step 1 + draft 1 + row 1) = ~600 tasks. With an advanced model, 5 per email = ~1,000 | Same | Pro 750 ($19.99 annual), or the 1,500 tier ($39 annual) with advanced models |
| Make | 2,880 credits for the checks + ~3 per email × 200 = ~3,480 credits | 720 + 600 = ~1,320 credits | Over Free's 1,000 either way, so Core ($9 annual, 10,000 credits) |
The pattern: n8n cost follows how often you run, Zapier cost follows how many items × steps, and Make cost follows both (cheaply). Double the email volume and the n8n bill stays the same. Zapier's and Make's counts grow with it.
Where each one fits
| If you… | Pick | Why |
|---|---|---|
| Want the easiest setup and the most app connections, low volume | Zapier | 9,000+ apps and the friendliest editor. Gets expensive per item once AI steps multiply |
| Want the lowest bill for simple app-to-app automations | Make | 10,000 credits for $9/month (annual) is hard to beat if each run has few modules |
| Run AI agents with many steps per item, or want to own the setup | n8n | Per-run billing, a full AI Agent node, code steps, and a free self-hosted option |
| Have data that must stay on your own server | n8n self-hosted | The only one of the three you can run entirely on your own infrastructure |
What this means for PlugStaff AI employees
PlugStaff AI employees ship as n8n workflow files plus a no-code version for a ChatGPT Custom GPT or a Claude Project. There is no Zapier or Make version, so on those tools you'd rebuild the logic yourself. The default schedules, so you can size your n8n plan honestly:
| AI employee | Default trigger (weekdays, business hours) | n8n executions / month |
|---|---|---|
| AI Customer Support Agent | Every 30 minutes, 08:00–18:30 | 440–506 |
| AI SDR (reply handler) | Hourly, 08:00–18:00 | 220–253 |
| AI Bookkeeper (receipt capture) | 09:00, 13:00 and 17:00 | 60–69 |
| AI Bookkeeper (weekly summary) | Mondays 08:00 | 4–5 |
| AI SDR prospecting, Content Marketer, Recruiter, Cold Email Research Agent | Run on demand (manual trigger) | 0 (manual runs from the editor don't count) |
| All five on Workforce | 724–833 (about 790) |
- On n8n Cloud Starter (2,500 executions): all five AI employees at their default schedules use about a third of the plan, leaving room for your own workflows.
- Want faster replies? Support every 15 minutes in business hours is about 1,000 a month, and the full Workforce then totals about 1,300: still under Starter. Every 15 minutes around the clock (2,880) is not.
- No n8n at all: use the Custom GPT / Claude version. It runs when you paste work in, with no automation bill.
Want to compare the cost of the AI employee itself with hiring? Try the ROI calculator or the Hire-or-Automate Scorecard. Building your own instead? Start with how to build an AI support agent in n8n. See also AI employees for small business.
FAQ
Is n8n cheaper than Zapier for AI agents?
Usually, once each item goes through several steps. n8n bills one execution per workflow run, however many steps it has (Cloud Starter €20/month annual for 2,500 runs, or self-hosted for no licence fee). Zapier bills every successful action step, and AI steps use 1–5 tasks each (Pro from $19.99/month annual for 750 tasks), as published in October 2026.
Is Make cheaper than Zapier?
For simple automations, yes: Make Core is $9/month billed annually for 10,000 credits, against Zapier Pro at $19.99/month annual for 750 tasks. But Make counts every module action, including each polling check, so frequent schedules use credits even when nothing new arrives.
Does a polling trigger use up my plan?
On n8n Cloud each scheduled run is one execution, even if it finds nothing. On Make each trigger check uses a credit. On Zapier, trigger checks don't use tasks; only successful action steps do.
Can I self-host to avoid the bill?
Only n8n offers a free self-hosted Community Edition. You then pay for the server and spend your own time on updates and backups. Zapier and Make are cloud-only for small businesses.
Do PlugStaff AI employees work on Zapier or Make?
No. They ship as n8n workflow files plus a no-code ChatGPT Custom GPT / Claude Project version. On Zapier or Make you would rebuild the logic yourself.
PlugStaff has an AI employee for each role above: $19/month for one, $49/month for all five. Browse AI employees.