GatwAI

n8n

n8n connects AI with business workflows: cloud or self-hosting, execution-based billing, data flows, limitations and two practical pilot workflows.

Provider sources checked

What does n8n do?

n8n orchestrates workflows that combine API calls, data transformations, language models and human decisions. A visual flow shows how data moves between systems, while developers can add custom logic. Its distinctive value lies in processes spanning several data sources and individual rules, rather than simply generating a piece of text. A pilot should cover the whole recurring process: input validation, processing, approval and failure handling. Cloud hosting reduces infrastructure work; self-hosting offers more control but requires updates, backups and monitoring. The inspectable software uses a fair-code model. That does not imply unrestricted permission for every commercial resale arrangement.
Editorial research · 2026-11-14How we research →

Who is this tool for?

  • Technical operations teams working across business systems.
  • Developers combining automation with their own infrastructure.

Strengths

  • Visual branches and custom code can coexist in one workflow.
  • Cloud and self-hosting support different operating models.
  • Billing complete workflow executions helps compare multi-step processes.

Limitations and requirements

  • Connectors, failure paths and permissions need maintenance.
  • Self-hosting transfers security and availability work to the team.
  • A technically successful workflow can still carry an incorrect model response.

Who is behind the product?

Company
n8n GmbH
Headquarters
Berlin, Germany.
Company founded
2019, by Jan Oberhauser according to the official press information.
Product launch
Not clearly established by the primary sources reviewed.

The founding information establishes company history; it is not treated as a separate first product-release date.

Suggested ways to use it

Suggested workflows or verification steps. These are not hands-on tests; for unverified products they are not a recommendation to use the service.

Triage support requests

  1. Prepare anonymized examples and a fixed category list.
  2. Combine model classification with required fields and an uncertainty route.
  3. Have staff confirm suggestions and log misclassification.

Goal: A proposed pilot measuring assistance with initial triage.

Review source changes

  1. Periodically retrieve permitted product pages and compare their content.
  2. Summarize changes with source URL and retrieval date.
  3. Require editorial approval before publishing anything.

Goal: A traceable review report rather than an unchecked automatic update.

Pricing, allowances and data

What should you check before choosing a plan?

Cloud Starter lists €20 per month billed annually for 2,500 executions; Pro lists €50 for 10,000. Steps within an execution are not the billing unit. Concurrency, log retention and administration differ. Self-hosting does not make the overall operation cost-free: infrastructure, maintenance and potentially model APIs remain. Check the applicable licence and overage arrangements before adoption.

What should you check about data and permissions?

The privacy policy excludes using personal data from services and integrations to train n8n or third-party ML models. The security page currently places cloud hosting on Azure in the EU. A connected model provider can still introduce its own data flows and terms. Self-hosting makes the team responsible for areas such as TLS, encryption, credentials and retention.

External reviews and available evidence

No GatwAI hands-on test or current independent comparison is included.

Reported positives

  • Documented hosting and billing options enable a concrete cost comparison.

Reported cautions

  • Provider examples do not establish the error rate in your process.

Which alternative fits which task?

Dify →

For a stronger focus on AI applications and knowledge retrieval.

Frequently asked questions

Is n8n entirely free?

No. Self-hosting, cloud subscriptions and external model costs must be considered separately.

Does EU hosting keep every data flow in the EU?

Not automatically. Every connected service and model call needs to be included in the data-flow assessment.