A free n8n workflow flags fake supplier invoices before anyone pays them — and Germany's e-invoice switch makes it a service bookkeepers can sell
Germany's e-invoice switch turned bank details into fields an agent can check. A builder published a free workflow that checks the IBAN, the VAT ID and BaFin warnings on every incoming invoice and flags red ones before payment. Bookkeepers could sell that as a monthly check.
Published Nov 9, 2026
The Wow
What someone built
An importable n8n workflow: parse the XRechnung or ZUGFeRD invoice, check the IBAN against the Bundesbank bank list, the VAT ID against the EU's VIES service and the seller against BaFin warnings, then log a traffic light to Google Sheets and alert Slack.
- Made by
- fast_data_asap (Fast Data) on DEV (not us)
- Original repository
- data-recipes recipe 8: e-invoice check workflow (MIT)
- License
- Workflow and README are MIT in the data-recipes repo; the actors it calls are paid per result
- What you need
- An n8n instance
- An Apify account and token
- Google Drive, Google Sheets and Slack (or SMTP for email alerts)
- Install
- Import workflow.json from the data-recipes repo folder 8-ai-finance-assistant-einvoice-check into n8n, add an Apify token as a Header Auth credential, then connect Drive, Sheets and Slack.
- A business you could build on it
- Pre-payment invoice check for bookkeeping offices
Why it's remarkable
The official KoSIT test invoice passes a plain IBAN checksum, but its bank code does not exist and VIES calls its VAT ID invalid. We checked both on Nov 9, 2026. A format check would have let it through.
The Opportunity
The business it reveals
Every German business must accept e-invoices now and issue them by 2028. A bookkeeping office that checks every client's incoming invoices before payment, and adds a known-IBAN rule per supplier, has a clear monthly service to sell.
Who would pay
German bookkeeping and tax offices handling invoices for small clients, then the small firms directly.
The Flip
3 ways you could turn this into a business
Pre-payment invoice check for bookkeeping offices
- Who pays
- German bookkeepers and Steuerberater offices handling incoming invoices for small clients
- How it would earn
- A monthly fee per client for the check, the known-IBAN list and a red-flag report (our hypothesis)
- Why it might work
- The e-invoice rules give every client the same structured data, and the per-invoice cost is reported at under two cents.
- What already exists
- Accounting and AP software that reads e-invoices; manual checks; the bank's payee-name check at transfer time.
Supplier bank-detail change desk
- Who pays
- Small manufacturers and wholesalers paying many suppliers
- How it would earn
- Setup plus monthly: every invoice whose IBAN differs from the supplier's known IBAN is held and verified by phone (our hypothesis)
- Why it might work
- The author names this as the fraud the base workflow misses and the first extension he would build.
- What already exists
- Bank Verification of Payee; in-house four-eyes checks.
E-invoice readiness setup for small firms
- Who pays
- Small firms that must issue e-invoices from 2027 or 2028
- How it would earn
- A fixed setup fee: inbox, parser, checks and an archive sheet, done in a day (our hypothesis)
- Why it might work
- Issuing becomes mandatory in 2027 for firms above 800,000 euros prior-year turnover and in 2028 for all.
- What already exists
- Accounting software vendors and tax offices offering their own onboarding.
Why we'd try "Pre-payment invoice check for bookkeeping offices" first: One bookkeeping office brings many clients at once, the data format is fixed by law, and each red invoice is a visible save.
Build It
How you could build it
The published n8n e-invoice check plus a supplier known-IBAN rule, one folder and sheet per client, and a weekly red-flag summary for the bookkeeper.
First step: Import the workflow, drop the KoSIT test invoice into the folder and confirm you get a red row, then add a supplier tab with one known IBAN and make a changed IBAN turn red.
Agents and tools
- n8n
- Apify actors for parsing and IBAN, VIES and BaFin checks
- Google Drive, Sheets and Slack
- ChatGPT, Claude, Gemini or Grok Bot to draft the weekly summary
Setup
- Import workflow.json and run the KoSIT test invoice until you see a red row.
- Add a supplier master-data tab and a rule: IBAN differs from the known IBAN turns red.
- Create one Drive folder and sheet per client; use the webhook with private links for real invoices.
- Have your assistant write a plain-language weekly summary of red and yellow invoices.
- Pilot with one client for a month and log every catch.
What the agent does
- Parses each invoice and runs the IBAN, VIES and BaFin checks
- Assigns the traffic light with reasons
- Drafts the weekly summary for the bookkeeper
What you do
- Sells to bookkeeping offices and sets up each client
- Keeps supplier bank details current
- Reviews red and yellow invoices and calls suppliers to confirm changes
Your first customer: A bookkeeping office in your network, offered a free one-month pilot on one client's incoming invoices.
Test demand first: Run the workflow on ten real (anonymized, client-approved) invoices from one office and show them the results. Success: they ask what it costs to run it on every client.
Your starter instructions
Paste this into the AI agent you use (ChatGPT, Claude, Gemini, Grok Bot or another) as your first message. Copying it starts work in your own agent; it doesn't create or deploy a business by itself.
Running it
What stays human: Selling to bookkeeping and tax offices, keeping each client's supplier IBAN list current, reviewing every red and yellow invoice, and calling the supplier on a known number when bank details change.
- Startup cost (estimate)
- Under $50 to run a pilot (estimate) (Estimate: n8n can be self-hosted for free, the workflow is free, and the author lists actor fees of $0.0145 per invoice (reported), about $1.45 per 100 invoices)
- Running cost
- About $1.45 per 100 invoices in actor fees plus review time (actor list prices; hours are an estimate) (According to the author's list prices: $0.003 parse, $0.0005 IBAN, $0.001 VAT ID and $0.01 BaFin screen per invoice; screening repeat suppliers monthly cuts it to $0.0045. Review hours are not measured)
- How much the agent runs
- Agents run one core part; people run the rest.
Pricing your own version? Use the free startup-cost worksheet.
The evidence
What the source shows: The author's DEV post (Nov 9, 2026), the MIT workflow and README in the data-recipes repo, the KoSIT test invoice, and our own VIES query. Checked Nov 9, 2026.
What's ours: The bookkeeper check, bank-detail change desk and e-invoice setup flips are our applications.
Worth planning around
- The author also sells the actors the workflow calls.
- It catches fake bank details, not a real supplier's invoice with a swapped IBAN, until you add a known-IBAN rule.
- It is a plausibility check, not tax advice.
Sources
- fast_data_asap, How to build an AI accounts-payable check for German e-invoices with n8n (DEV, Nov 9, 2026)
- data-recipes repo, recipe 8: e-invoice check workflow and README (MIT, checked Nov 9, 2026)
- KoSIT XRechnung test suite, business case 01.01a (checked Nov 9, 2026)
- German Federal Ministry of Finance, FAQ on the mandatory e-invoice (checked Nov 9, 2026)
- Bavarian State Tax Office, E-Rechnung timeline (checked Nov 9, 2026)
Sources reopened Nov 9, 2026. Spot a mistake? Tell us and we'll correct it in public.
Keep reading
- How people make money with n8n (guide)
- How much of the work do AI agents actually run? (guide)
- He still can't write code. His five-person ad agency now runs on about 270 tools he built with Claude. Every owner-run firm needs its first five
- A marketer who doesn't code built an elk-call coaching app with Claude Code and says it hit a $650K run rate in 110 days. Every skill learned by sound is the same opening
- Marc Lou's agent chased a broken-shoe refund through three support desks and got his 850 AED back. Small stores now need a claims desk that can answer agents
- levelsio says Claude Code can design furniture part by part and hand you the parts list. A non-designer made a table. Local carpenters could build what people design with agents
- Every discovery and launch plan