How to use n8n as an automation layer for business software

Your team enters an order into FileMaker. Then someone opens Exact Online and types the same order in again — every single order, every single day. Meanwhile your CRM has no idea the deal closed, so the account manager follows up on a customer who already received their invoice. This is what it looks like when business software grows faster than the glue holding it together. This article shows you exactly how n8n works as an automation layer between those systems, what that looks like in practice, and how to decide whether it belongs in your stack.

What is n8n, and why does it matter for business software?

n8n (pronounced "n-eight-n") is an open-source workflow automation platform. Think of it as a programmable switchboard that sits between your applications and routes data between them based on rules you define. Unlike Zapier or Make, n8n can be self-hosted on your own server or cloud environment — which matters enormously for companies that deal with sensitive business data and can't afford to route customer records or financial transactions through a third-party SaaS platform they don't control.

Where n8n stands apart:

• Self-hosted or cloud — you own the execution environment and the data.
• Fair-code license — free for internal use, with a commercial license only required if you resell it.
• Native code nodes — you can drop into JavaScript or Python mid-workflow when no-code nodes aren't enough.
• 400+ integrations — including REST APIs, webhooks, databases, and AI model providers.
• Visual workflow editor — non-developers can read and understand what a workflow does, even if a developer built it.

For SMEs running FileMaker, an ERP like Exact Online, and a CRM like HubSpot or Salesforce, n8n is the connective tissue that turns three isolated islands into one coherent data ecosystem.

Why manual re-entry is more expensive than it looks

The visible cost of manual data entry is the time: 10 minutes per order, 20 orders a day, one employee — that's roughly 35 hours a month of pure copy-paste work. But the hidden costs are worse:

• Errors that compound. A customer name spelled differently in two systems means duplicate records, broken reporting, and invoices that don't match.
• Delayed decisions. If your FileMaker production data only reaches your ERP at end-of-day (or end-of-week), your financial overview is always running behind reality.
• Integration debt. The longer systems stay disconnected, the more workarounds accumulate — spreadsheets, shared inboxes, Slack messages used as a makeshift database.

A concrete example: a wholesale distributor processes 80 orders per day. Each order is confirmed in their FileMaker order management system, then manually posted to Exact Online for invoicing. At 8 minutes per order, that's over 10 hours of daily admin. One missed posting means a delayed invoice, a disrupted cash flow, and an awkward customer conversation. n8n eliminates that gap entirely.

How does n8n actually connect FileMaker and Exact Online?

FileMaker exposes its data through the FileMaker Data API — a REST-based interface that lets external systems read records, create new ones, run scripts, and trigger layouts. Exact Online has its own REST API. n8n can speak to both natively.

Here's how a real FileMaker → Exact Online order sync workflow is structured:

1. Trigger: A webhook or scheduled poll detects a new or updated order record in FileMaker (e.g., status changed to "Confirmed").
2. Fetch: n8n calls the FileMaker Data API to retrieve the full order record — line items, customer ID, delivery address, amounts.
3. Transform: A Function node maps FileMaker field names to Exact Online's expected data structure. This is where you handle edge cases: missing VAT codes, unknown debtor IDs, rounding differences.
4. Lookup: n8n queries Exact Online to check whether the customer (debtor) already exists. If not, it creates one.
5. Post: n8n creates a sales order or invoice in Exact Online via its API.
6. Write back: n8n updates the FileMaker record with the Exact Online document number and a sync timestamp — so both systems confirm the exchange happened.
7. Error handling: If any step fails, n8n sends an alert to a Slack channel or email with the exact record ID and error message, so a human can intervene precisely — not hunt blindly.

This entire workflow runs in seconds. The employee who used to re-type orders never touches this data again. They see the Exact Online reference appear in FileMaker automatically.

How do you add AI to an n8n workflow?

One of n8n's most powerful — and underused — features is its native support for AI model APIs, including OpenAI, Anthropic, and local LLMs via Ollama. This makes it possible to insert an AI reasoning step inside an otherwise standard business automation.

Practical examples of AI inside an n8n workflow:

• Incoming email triage: A customer emails your support address. n8n intercepts it, sends the text to an OpenAI node, classifies it as "complaint," "order," or "question," and routes it to the right team queue — or creates a CRM ticket automatically with a suggested priority level.
• Contract summarization: A PDF lands in a shared folder. n8n extracts the text, sends it to an LLM, and writes a structured summary (parties, key dates, renewal clause) back into a FileMaker record — cutting document review time from 20 minutes to 20 seconds.
• Anomaly detection in orders: n8n compares each incoming order against historical averages. If an order quantity is 10× the norm, the AI node flags it for human review before it ever reaches Exact Online.
• AI-assisted quote generation: A salesperson fills in a few fields in FileMaker. n8n picks up the record, sends the product specs and customer history to an LLM, generates a personalized proposal draft, and creates a Google Doc or Word file — ready for final review.

These aren't science-fiction workflows. They're running in production today, built with n8n's AI Agent nodes and standard API calls. The key insight is that AI doesn't replace the automation layer — it becomes one step inside it, just like a database lookup or a data transformation.

n8n vs. Zapier vs. Make: which one should you use?

This is the question every IT manager asks. Here is an honest, non-marketing comparison:

	Zapier	Make (Integromat)	n8n
Hosting	Cloud only	Cloud only	Self-hosted or cloud
Data ownership	Vendor's servers	Vendor's servers	Your own server
Pricing model	Per task/month	Per operation/month	Free (self-hosted); flat fee (cloud)
Custom code	Very limited	Limited (JSON/JS in some nodes)	Full JavaScript & Python nodes
Complex logic	Workarounds needed	Moderate	Native (branches, loops, error paths)
FileMaker Data API	No native node	No native node	HTTP Request node (full control)
AI / LLM nodes	Basic	Basic	Native (OpenAI, Anthropic, Ollama)
Best for	Simple, fast setups	Visual, moderate complexity	Complex, data-sensitive, scalable flows

When Zapier still wins: You need something running in 20 minutes with zero server knowledge and the data isn't sensitive. Connecting a Typeform to a Mailchimp list — Zapier is fine.

When Make wins: You want a visual, flowchart-style tool with good error visibility and moderate complexity, without managing infrastructure.

When n8n wins: You're handling financial data, customer PII, or healthcare records. You need custom logic that no-code tools can't express. You're tired of paying per-operation fees that scale unpredictably. You want AI steps embedded in business workflows. You have a developer (or a partner) who can set it up and maintain it.

For most SMEs with a FileMaker backbone and an ERP integration need, n8n is the right choice — not because it's fashionable, but because the per-operation pricing of Zapier and Make becomes genuinely painful at business scale, and the data routing through third-party servers creates compliance risk that most companies only discover after a problem.

What does a production-ready n8n setup look like?

Running n8n in a demo takes 10 minutes. Running it reliably in production is a different conversation. Here's what a serious setup requires:

Infrastructure checklist

• n8n installed on a dedicated VPS or container (Docker is the standard)
• PostgreSQL database as the backend (not the default SQLite — it won't scale)
• Reverse proxy (nginx or Caddy) with SSL termination
• Environment variables for all credentials — never hardcoded in workflows
• Regular automated backups of the n8n database and workflow exports
• Monitoring and uptime alerts (e.g., UptimeRobot, Grafana)

Workflow design checklist

• Every workflow has error handling — a failing node sends an alert, not silence
• Sensitive data (API keys, tokens) stored in n8n's built-in credential vault, not in node parameters
• Workflows are version-controlled (export as JSON, commit to Git)
• Each workflow does one thing — don't build a 40-node monolith
• Test with production-like data before going live
• Document what each workflow does and who owns it

Skipping these steps is exactly how companies end up with a "working" automation that silently fails for three weeks before someone notices an order never reached the ERP.

What are the real gotchas when connecting FileMaker via n8n?

Anyone who has actually built FileMaker ↔ n8n integrations will have hit these:

1. FileMaker Data API sessions expire. The API uses session tokens that time out after 15 minutes of inactivity. Your n8n workflow needs to handle re-authentication — either by logging in fresh at the start of each execution or by catching the 401 error and retrying.

2. FileMaker field names are case-sensitive and layout-dependent. The Data API only exposes fields that exist on the specific layout you call. If a field isn't on that layout, it won't appear in the response — and no error tells you why.

3. Exact Online API rate limits are real. Exact Online enforces per-minute and per-day API call limits. If you're syncing hundreds of records, you need throttling logic in n8n (the "Wait" node or a batching strategy) or you'll hit a 429 and your workflow will stall.

4. Data type mismatches. FileMaker timestamps look different from Exact Online date formats. Exact Online expects amounts as numbers; FileMaker sometimes stores them as formatted strings. Build explicit transformation nodes — never assume the format will match.

5. Idempotency. What happens if the workflow runs twice for the same record? Without a deduplication check (e.g., "does this Exact Online order already exist?"), you'll create duplicate invoices. Always design for the workflow running more than once.

Frequently asked questions

Do I need a developer to use n8n? For simple workflows (webhook → transform → API call), a technically capable operations person can learn n8n in a few days. For complex integrations involving custom authentication, error-handling logic, or database queries, a developer will save you weeks of frustration and prevent silent failures in production.

Can n8n replace our iPaaS subscription entirely? For many SMEs: yes. If you're currently paying €200–€800/month to Zapier or Make for business-critical workflows, a self-hosted n8n instance typically costs €20–€50/month in server costs — with no per-operation fees. The break-even is usually within 2–3 months.

Is n8n secure enough for financial data? Self-hosted n8n is as secure as your infrastructure. Because data never leaves your server, it's generally more compliant than routing financial transactions through a third-party SaaS. You control encryption, access logs, and data retention.

What happens if n8n goes down? Workflows don't execute. Webhooks queue or drop, depending on configuration. This is why production monitoring, uptime alerts, and retry logic matter — see the infrastructure checklist above. For mission-critical flows, a redundant setup or a managed n8n cloud subscription is worth considering.

Can n8n work with systems other than FileMaker and Exact Online? Yes. n8n has native nodes for HubSpot, Salesforce, Shopify, WooCommerce, Microsoft 365, Google Workspace, Slack, and hundreds more. Any system with a REST API or webhook can be connected via the HTTP Request node.

How long does it take to build a FileMaker ↔ Exact Online sync? A straightforward one-directional order sync (FileMaker → Exact Online) typically takes 2–4 days of development for an experienced integrator — including testing, error handling, and write-back confirmation. A bidirectional sync with conflict resolution takes longer.

---

If your systems are siloed and your team is spending real hours every week bridging the gap by hand, the integration architecture question is worth solving properly — not patching with another spreadsheet. Loggix designs and builds these automation layers as part of tailored business software projects: connecting FileMaker with ERP systems like Exact Online, embedding AI steps into operational workflows, and setting up n8n environments that are production-ready from day one. If you're ready to map out what a connected system would look like for your business, that's exactly the kind of conversation we have.