The Freelancer Who Built Your n8n Workflow Is Your Biggest Risk

The workflow that routes your leads is running right now. You do not know exactly how. One person built it, and that person is no longer answering.

When it breaks, and a fifty-node n8n workflow with no error handling will break, you will not get an alert. Leads will simply stop arriving. You will notice a slow week, then a slower one, and by the time someone traces it back to a silent failure in a workflow nobody can read, you have lost deals you never knew existed.

Who I am

I started with plain Python and plain JavaScript, writing automations at university for my own projects, first scripts that processed data and called APIs. Then Make.com, Zapier, n8n. I have built with all of these tools and run them in production, for my own projects and for clients.

At some point the no-code tools hit their ceiling. Too complex. Too expensive. Too little control. I went back to code, and since then I have built custom automation systems for companies: webhook layers, job queues, monitoring dashboards. If you want to see what "I know this deeply" means in practice, I spent three months in the weeds of one platform. This article is the business version of what that taught me.

I mention this so you know I am not someone who reads comparison articles and summarizes them. I have run these systems myself, and I know exactly which problems appear when you pick the wrong tool, or the wrong person, for the job.

The real problem is not n8n

n8n is a good tool. Make.com is a good tool. So is Zapier.

The problem is what happens when a freelancer operates these tools for you, and you cannot read what they built.

You pay for the workflow, but you do not own it in any way that matters.

An n8n workflow with 40 nodes looks like technical documentation. Whoever built it understands it. Whoever inherits it, a new freelancer, an employee, or you, faces a diagram nobody can explain. Every node carries context recorded nowhere. Every if-condition follows logic that lived in one person's head and left when they did.

This creates three costs, and only one of them shows up on an invoice.

The dependency cost

For every small change, a new email template, an extra field, a changed endpoint, you wait on one person to have time, to reply, and to be skilled enough to do it cleanly. You are the buyer in a market where the seller controls the knowledge. You set neither the timeline nor the price.

The handover cost

When the freelancer leaves, the next one needs three to five hours just to understand what the workflow does before changing a single thing. You pay for that archaeology. Then you pay again for the change itself, which takes longer than it should because the new person does not have the context the old one had.

The silent-failure cost

This is the one nobody prices in, and it is the largest. A workflow built without error handling fails quietly. No alert. No log anyone reads. The lead routing stops, the data sync drifts, the follow-up emails stop sending, and the business runs blind until someone notices a number trending down. By then the damage is done and undiagnosable.

You have invested thousands, the workflow half-works, you are afraid to touch it, and the thing routing your revenue depends on someone who may not pick up the phone next week.

If you run cold email or any outbound, this is not hypothetical, it is the whole game. Your enrichment workflows, the ones that research each prospect and write the personalized line, are the difference between replies and silence. A campaign to a few thousand prospects is tens of thousands of automated steps: scrape the site, find the email, verify it, score the company, write the opener. When that pipeline fails quietly, the campaign does not stop. It keeps sending, but generic. Thousands of prospects get a flat, templated email. Your reply rate collapses, your domain reputation takes the hit, and the list you paid to build is burned, all from a workflow failure nobody saw until the numbers came in.

What changed, and why this is the right moment

Claude Code changed the relationship between a company and its automation. Not because it makes automation trivial, but because it changes where the dependency sits.

Here is the honest version, because the dishonest version is everywhere. Claude Code does not make you an engineer overnight, and someone still has to set up the foundation correctly the first time. But once that foundation exists, the day-to-day changes completely.

The old loop: workflow change — contact freelancer — wait for a quote — approve — wait for delivery — review — request a fix.

The new loop: workflow change — describe in plain language what should happen — Claude Code writes it — test — done.

The point is not speed for its own sake. It is that maintaining and extending your workflows no longer requires the one person who originally built them. Anyone on your team who can clearly describe a process can change a workflow built this way. Today. Without waiting on anyone.

"When a new lead form is submitted, send an email to sales, create the contact in HubSpot, and post a Slack message to the responsible account manager."

That is the level of description Claude Code works from. The black box becomes readable text that explains itself.

What it actually costs: the freelancer, and the alternative

The numbers below are illustrative ranges based on typical market rates, not a fixed quote. But the shape is what matters.

What Windmill is, and why it beats n8n here

Windmill is an open-source workflow platform. You write short scripts in Python or JavaScript, or you let Claude Code write them, and you get automatically: a webhook endpoint for every workflow, a dashboard showing what is running and what failed and why, logging for every job, plus scheduling, retries and failure notifications.

The difference from n8n is the one that matters for your risk. A Windmill script is readable text. You open it and ask Claude Code "what does this do?" and get a clear explanation in 30 seconds. An n8n workflow with 40 nodes cannot be read and explained that way, by Claude or by your next hire.

Windmill is not a black box. No new employee, no new freelancer has to "read into it." And because it has real error handling and a real dashboard, failures surface instead of hiding.

You are no longer dependent on whoever built it. Anyone who can describe a process can maintain it. Today. No lead time. No one to chase.

Which solution for which case

Few simple workflows, low volume, no growth planned: n8n, Make or Zapier are enough. The overhead of Windmill is not worth it. Just keep the workflows simple and documented.

You are growing, the workflows touch revenue, or you already cannot read what you have: Windmill self-hosted with Claude Code. A €35 server per month, no execution limits, no vendor lock-in, and your workflows are readable by your own team.

You run cold email or outbound at volume: This is the case where the infrastructure and the campaign are the same decision. Personalizing thousands of prospects means tens of thousands of enrichment executions, and the pipeline running them has to be both reliable and readable. Get it right and you can personalize at scale; get it wrong and you are either sending generic email or paying a fortune per execution to managed infrastructure. This is the work I do most.

You want maximum simplicity without a server: Windmill Cloud or Trigger.dev Cloud. You pay per execution, cheaper at low volume, more expensive than self-hosting at high volume.

What this really means for your business

This is not about which tool is technically better. It is about who can see inside the systems that run your revenue, and how fast you find out when one breaks.

Today, a freelancer holds knowledge you do not have, decides when they reply, and sets what they charge. The systems routing your leads are opaque to you, and they fail silently. With Claude Code and Windmill, you describe what should happen, the system is readable by your own team, and failures show up on a dashboard the same day. That is not a claim about software. It is a change in who holds control of your business.

Three things you can do today

1. Find your single points of failure. Are there workflows only one person understands, that you would not touch without asking the original builder? Those are the ones routing your revenue with no safety net. Take inventory first.
2. Test Claude Code on one existing workflow. Open it, describe it in three sentences, and ask Claude Code to write it as a Windmill script. You will see how small the migration is, and how much more readable the result is than the visual flow.
3. Price your real exposure, not just the fees. Add up the freelancer hours you paid for automation in the last 12 months. Then ask the harder question: if your lead-routing workflow failed silently for two weeks, what would that cost? That second number is the one that should drive the decision.

FAQ