Process automation with n8n -
self-hosted, privacy-oriented, no per-execution licence fees.

Process automation means letting software execute recurring, rule-based work instead of people: a trigger fires (a form submission, a webhook, a schedule, a new record), logic decides what happens, and one or more systems are written to. n8n is the tool we reach for first, because it covers the vast majority of these flows visually and can be self-hosted. Where a process needs more than a workflow engine - high-throughput data processing, custom APIs, event-driven services - we build it in code and connect it to the same setup. The tool is a decision, not the point: the point is that the process runs without a human in the loop and belongs to you.

n8n is a workflow automation platform that lets you visually connect apps, APIs, databases and AI models into automated flows - by drag-and-drop, without coding every step. Unlike pure no-code tools, you can extend any node with your own JavaScript or Python. It is used for routing leads into a CRM, data enrichment, document and invoice workflows, internal IT and ops automations, and AI agents. The key point: n8n can be self-hosted, so all data stays inside your own infrastructure.

Then we do not force it. n8n is excellent at orchestration: triggers, branching, API calls, retries, human-in-the-loop steps. It is not the right answer for everything - very high throughput data processing, latency-critical services or deeply custom business logic are better served by dedicated code. We build those in Python, Rust or Node, put an event bus such as Kafka or NATS in front where it helps, and expose a unified API layer. Both worlds run on the same infrastructure, and you own all of it.

The biggest difference is the pricing model. n8n counts one complete workflow run as a single execution - no matter how many steps it contains. Zapier charges per task, meaning per individual step. A workflow with 10 nodes that runs 10,000 times per month is 10,000 executions on n8n versus 100,000 tasks on Zapier - a documented cost gap of around 90 percent. Make sits in between (operations per module) but, like Zapier, remains a pure US SaaS with no self-hosting. n8n is the only one of the three you can run entirely on your own infrastructure - with full data sovereignty and no volume limits.

At moderate volume, Zapier and Make land around $600 to $900 per month. Over three years that is $21,600 to $32,400 - and at the end of it you own nothing: no code, no workflows you can move, no infrastructure. A self-hosted setup is a one-time build plus hosting, typically well under $100 a month for the server. The workflows, the code and the documentation are yours from day one.

At typical workflow volume (50,000 to 100,000 executions per month) the build pays for itself within 12 to 18 months, after which the system runs at hosting cost. The time saving usually arrives first: a single automated process that gives back 20 hours a month is worth around $1,440 a month at $72/h opportunity cost - roughly $17,280 a year. We calculate this for your actual processes before you commit, so the decision rests on numbers.

n8n itself is only a tool - compliance comes from how it is operated. With self-hosted n8n on an EU or German server, customer data, API keys and webhook payloads never leave your infrastructure, and no data processing agreement (DPA) with a US SaaS provider is required. That is exactly the advantage over cloud automators that transfer data to the US. We host n8n on German infrastructure (Hetzner) and work to ISO/IEC 27001:2022 - so the operation is set up in a privacy-friendly way. This is a technical assessment and does not replace legal advice.

It removes an entire class of risk: your data never reaches a third-party server, and no vendor holds access to your business data. Workflows run in Hetzner's ISO 27001 certified data centers in Germany or on hosting you choose, with TLS 1.3 in transit, audit trails and backups. You control every data point and every credential. What you take on in exchange is operational responsibility - which is why we offer to run updates, monitoring and backups for you.

Yes. The n8n Community Edition is source-available and free to use - when self-hosting you pay only for the server, typically around 5 to 20 EUR per month on a small VPS. For comparison: n8n Cloud starts at 20 EUR/month (2,500 executions) and rises to 667 EUR/month (Business, 40,000 executions); Zapier plans climb quickly into four- and five-figure monthly bills at scale. Because self-hosting has no execution limits, costs do not grow with volume. We do not quote flat project prices - every project is scoped individually.

No, and the distinction matters: n8n is not open source in the strict sense, but fair-code under the Sustainable Use License. That means source-available - the code is visible and free for internal business or personal use. There are no license fees for internal use within your own company. As soon as n8n access is passed on to third parties or offered as a hosted service, a commercial license from n8n is required. For our work that means: we build and maintain n8n on your own instance, which belongs to you. We do not host client access inside a central FW Delta instance, because that is exactly what would require a commercial licence.

Anything recurring, rule-based and cross-system: capturing leads from forms and routing them into a CRM, lead enrichment and scoring, CRM data hygiene (duplicates, normalization), drip and nurture triggers, invoice and document workflows, order processing, syncing between tools, scheduled reports, and internal ops and IT automations. The three most expensive manual processes in most companies are moving data between systems (15 to 25 hours a month), sending standard communication by hand (10 to 20 hours) and rebuilding reports (8 to 15 hours). For scale: Delivery Hero saves over 200 hours of IT work per month with a single n8n workflow, and StepStone integrates new data sources around 25 times faster.

Yes. n8n ships over 70 AI-specific, LangChain-based nodes and supports AI agents, RAG pipelines and vector databases (Pinecone, Qdrant, Weaviate, Chroma) plus more than 12 LLM providers. Through Ollama you can run local language models - so a complete LLM pipeline runs on your own server without data flowing to external APIs such as OpenAI. That is sovereign AI: automation with the intelligence of a language model, but under full data control.

We do, if you want. Managed n8n with us means: we set up the server (Hetzner Germany), apply updates and security patches, monitor the workflows, run backups and step in when something breaks. Unlike a SaaS subscription, though, the solution belongs to you - you get full access, the workflows and the documentation. You can take over operations yourself at any time or move to a different provider. No lock-in.

You own the code and the workflows, so you can add whatever you need - yourself, with your team, with any developer you choose, or with us. There is no vendor gating features behind a plan and no technical ceiling on how many workflows or executions you run. Maintenance and build-out packages are available, but nothing about the setup requires them.

No. The system runs on its own: error handling, retry logic, monitoring and alerting are part of the build, not an afterthought. You get complete documentation and, where it makes sense, a management view of the running workflows. Most clients never touch the workflow editor after handover - and the ones who want to can, because the setup is documented and open to them.

In two cases almost always: when your automation volume grows (task-based Zapier/Make costs then explode while self-hosted n8n stays flat) and when data protection matters (EU self-hosting wins back your data sovereignty). For small, static setups with a few hundred runs per month, an existing SaaS plan may be enough. We migrate existing Zapier and Make scenarios to n8n and calculate the savings over 1 to 3 years up front, so the switch is based on numbers, not a hunch.

In five steps: first analysis - we map your processes and identify the workflows with the highest leverage. Second build - we build the workflows on your infrastructure. Third test - each flow is verified with real data and failure cases. Fourth go-live - a controlled production start with monitoring. Fifth handover - you receive access, documentation and, on request, training. Simple setups go live within days, more complex integrations take a few weeks.