Skip to main content

Work that gets itself done.

Process automation you can watch: in the Automation Showroom, 20 workflows from finance, sales, service and HR run as demos, step by step. Green means software does it, amber means a person decides.

Log

demo, simulated

  1. Discharge LetterDischarge Triggereddone
  2. AI-Powered Lead QualificationLead Captureddone
  3. Employee OnboardingContract Signedapproved
  4. DunningDaily Rundone
  5. Invoice VerificationIntake Channeldone

How an automated workflow runs.

Whatever follows the same rules every time is done by software. Where judgement is needed, the workflow stops and puts everything in front of a person, ready to decide.

  1. Intake Channel
  2. Data Extraction
  3. PO Lookup
  4. Clarification Workflow
  5. AI Account Suggestion
  6. Approval Cascade (a person decides here)
  7. Discount Alert
12→2 daysThroughput · DemoSimulation: 0 cases completed while you watchSee this workflow
  1. Trigger

    An invoice arrives by email, a contract is signed, a customer writes in. That starts the workflow; nobody has to remember it.

  2. Steps

    Software does what is the same every time: read, match, enter, notify.

  3. Decision

    Where judgement is needed, the workflow waits for a person. Approving then takes seconds instead of minutes.

  4. Result

    Booked, sent or filed, with every step logged. Traceable, even a year later.

The whole company, connected.

A single workflow saves minutes. It really shows once the workflows interlock: the order triggers the delivery, the delivery the invoice, the unpaid invoice the reminders. Nobody carries information from department to department any more.

Connections in the example company

  • A request comes in, is sorted by topic and urgency and lands with the right person in sales.

    See it as a demo

20 workflows to look at.

From finance, sales, service and HR. Each one shows what used to be manual, how it runs now and where a person decides.

Demos with modelled figures, not client projects.

A receptionist at a doctor's practice talks on a headset while looking at her screen.
At the practice: callbacks, appointments and doctor's letters while the next patient is already at the desk.
A master craftsman at a laptop in his workshop office, timber and tools behind him.
In the trades: quote, order and invoice from the same data instead of typed three times.
A car mechanic checks a work order on a tablet in the workshop.
In the workshop: the work order on a tablet instead of a note on the wing.

Staged scenes, not client projects.

What it brings, and where it does not.

In the 20 demos, 180 steps run on their own: read, match, book, notify. At 19 points a person decides, on purpose.

Automation happens in what is already there. Connected in the demos: DATEV, SAP, Odoo, Lexware Office, sevDesk, Personio, HubSpot, Shopify, Stripe, Zendesk, Jira, Confluence, Slack, Microsoft Teams, SharePoint, Google Workspace, DocuSign, Peppol, n8n, Make.com.

I am not a certified partner of these vendors.

Where automation does not help

  • When the process runs differently every time.

    Then rules are missing, and someone has to set them first. Without a standard there is nothing a machine could repeat.

  • When a step is actually unnecessary.

    Better to drop it than to speed it up. An automated detour is still a detour.

  • When nobody is responsible.

    An automation without an owner lasts until the first exception. After that nobody knows why it does what it does.

The limits of automation
A business owner alone in the office in the evening, rubbing her temples in front of a laptop and stacks of paper.

Staged scene

Tuesday, 7:40 pm

Today's invoices are not booked yet.

Typing, chasing, searching: most of it nobody notices any more. It can still be measured.

Where things get stuck, the people who work with it every day know.

So you do not start with a tool but with collecting: note disruptions for four weeks, convert them into euros, then drop what nobody needs. What remains gets automated.

The example is invented: 188 reports from four weeks, each dot is one of them.

Report

Reporting a disruption takes ten seconds.

Your team taps in what got stuck: by link or QR code, anonymously, without an account. No workshops, no interviews.

  • Sales: Asked production for a delivery date, no answer · 25 min
  • Production: Order without tolerances, sent back to sales · 40 min
  • Accounting: Reminder sent to a customer who paid long ago · 20 min

You can also start on your own: with the free disruption diary for iPhone and Android you note what slows you down and see a projection after one week.

Measure

Opinions become a number.

In the example, 188 reports come in over four weeks, sorted by cause and converted into euros.

  • Hand-over sales → production71 reports, 38 %
  • Entered twice41 reports, 22 %
  • Details missing34 reports, 18 %
  • Looking for information25 reports, 13 %
  • Other17 reports, 9 %

€72,450per year

31.5 hours a week × 46 weeks × €50. Invented example.

Remove

Over half disappears without software.

104 of the 188 disruptions go away because a step is dropped, a field becomes mandatory or a responsibility is settled. That costs no licence, and nobody has to wait for it.

By cause
  • Hand-over sales → production40 of 71 removed
  • Entered twice28 of 41 removed
  • Details missing16 of 34 removed
  • Looking for information14 of 25 removed
  • Other6 of 17 removed

What a change brings can be played through first: the Flowrefy method models the process, FlowVisual runs it hundreds of times and shows where it costs most.

Automate

And the rest runs on its own.

Like the workflows in the showroom, only with your own systems and rules. After a few months you measure again at the same point to see whether the number has moved.

The method in detail
Three colleagues laughing over coffee in a bright office kitchen.

Staged scene

Wednesday, 10:15 am

Yesterday's invoices are booked. Nobody typed them.

Where a person has to decide, a person still decides. Only the typing, forwarding and chasing runs on its own.

What companies want to know first.

What can be automated at all?

Anything that regularly runs by the same rules: checking invoices, writing payment reminders, answering and routing requests, setting up new colleagues, approving documents. Where judgement is needed, a person stays in the workflow and gets everything prepared.

Do we need new software for this?

Usually not. An automation connects the systems you already have, such as DATEV, SAP, Odoo, Personio, your inbox or Excel. It is built with tools like n8n or Make.com and, where it fits, with AI models.

What is the difference between automation and AI?

An automation follows fixed rules: when this happens, do that. AI comes in where rules are not enough, for example reading a scanned invoice or classifying a customer request. In most showroom workflows AI is a single step, not the whole thing.

Are the showroom cases real client projects?

No. They are demos with modelled figures. They show the workflow, the tools involved and every point where a person decides, so you can see what something like it could look like for you.

Where is the best place to start?

With a workflow that happens often, always runs in a similar way and annoys your team. Incoming invoices, payment reminders and onboarding are typical candidates. Before that it pays to check which step can be dropped entirely, because what is gone does not need automating.

Which workflow annoys you the most?

Have a look in the showroom for something similar. Or place it in the self-check: about 5 minutes, free and without sign-up. You see the short result straight away, no email address needed.

Rather talk it through? Write to me