“Automate it” covers everything from a five-minute Zapier connection to a six-figure enterprise platform. Here's the honest map of what connects your tools, what it really costs, and what you can run yourself.
Automation tools fall into roughly three tiers, plus a heavyweight enterprise layer. Picking the right tier matters more than picking the right brand — most people overspend by reaching for a tier above what they need.
Pay a monthly fee, build in the browser, ship today. Starter plans run roughly €20–50/month and climb into the low hundreds as volume grows — you're metered by tasks, operations or credits, and you can't take the workflow elsewhere.
Run it on your own box — a small VPS or AWS EC2 instance, about €5–15/month — and the per-task meter disappears: you pay for the server, not the volume. The trade is a real one, though: you take on the updates, the backups and the security, and when it breaks at midnight nobody's on call but you. The licences also differ — some are fully permissive, others (n8n, Windmill, Automatisch) restrict reselling the tool itself as a service, which for your own workflows never bites.
Two more worth knowing if you self-host: Huginn, a long-running “agents that watch the web” system (MIT), and Automatisch, an open Zapier clone (AGPL-3.0).
When automations become long-running, mission-critical or need to survive failures, you move from no-code into code-first engines. This is engineering territory, not a weekend project. And if you're already on AWS, its native pieces — Step Functions, EventBridge, Lambda — do the same job inside your account.
Apache Airflow often comes up here — but it's a data-pipeline scheduler for engineers, not a no-code app-automation tool. Useful, just a different kind of tool than the rest here.
There's a heavyweight tier built for large organisations: deep governance, hundreds of connectors, and sales-led pricing that's almost never published — realistically €10k to €300k+ a year on annual contracts. For a solo operator or small business this is overkill and overpriced, but it's worth knowing the names so you recognise them in a tender.
Others in this bracket: SnapLogic, Microsoft's Power Automate, Informatica and Jitterbit. All sales-led, all enterprise-priced.
Don't start with the biggest project. Start with the dullest thing you repeat every week — logging enquiries, copying data between two apps, chasing reminders, generating invoices, triaging customer-service messages. Build that one cleanly so it holds, then expand. And for plenty of automations you don't need AI at all yet — fixed rules are cheaper and more reliable. AI is worth adding only when a step needs judgement that rules can't express — reading a messy email and deciding where it should go, say. That's when an AI agent sits at the core of the automation rather than beside it.
Where I fit. I usually build on n8n plus your APIs, with an AI core only where it pays for itself — one process at a time, fixed price per process. If you'd rather DIY, n8n or Activepieces is the place to start, and I'm glad to point you at the right first workflow on a call.
Zapier if you want the fastest possible start and don't mind paying per task. Make for more complex logic at a lower per-run cost. n8n if you'd rather self-host and stop paying per task as volume grows.
Yes — that's the main reason to. Tools like n8n, Activepieces or Node-RED run on a small server (~€5–15/month) for a flat cost, no matter how often your workflows fire. The trade: you take on the setup, updates and backups yourself, and you're the one on call when something breaks.
Only when a step needs judgement that fixed rules can't express — reading an unstructured email, drafting a tailored reply, deciding where something should go. For straight “when X, do Y” work, plain automation is cheaper and more reliable.
These are serious platforms, but built and priced for large companies — typically five to six figures a year. For a small business they're overkill; one of the tier-1 or tier-2 tools above will do the same job for a fraction of the cost.