Skip to content

Automation

The work a person shouldn't be doing.

Automating isn't installing a tool. It's understanding a process well enough to write down its rules, including the exceptions nobody mentions until the third meeting.

From €450

3-5 days

What gets automated

Tasks that meet three conditions.

Not all repetitive work is worth automating. We look for tasks that happen often, follow the same steps every time, and whose data is already digital. When all three are missing, the process itself usually needs fixing before any tool gets involved.

It repeats

Daily or weekly. Something that happens twice a year rarely justifies the maintenance.

It follows rules

If two people do it differently, someone first has to decide which way is the right one.

The data exists

In a spreadsheet, an inbox or an ERP. If it's on paper or in somebody's head, that's the first project.

Order intakeDocument creationQuotesInvoicingReport generationDatabase updatesNotifications and follow-upInternal operations

What it looks like inside

An automation, explained without the jargon.

It is nearly always the same idea: something arrives, it gets checked, it lands where it belongs and someone gets told. The rest is implementation detail.

These three are the patterns we repeat most. The tools at either end change; the shape doesn't.

Orders that arrive by email

Email with orderExtract dataValidateERPConfirmation

Documents that end up in the CRM

PDFDataCRM

Quotes from a web form

FormQuoteEmailDatabase

The validate step is the one other people's demos leave out and the one that decides whether the system holds. When something doesn't add up, it stops and tells a person.

How we find them

We watch the process where it happens.

We sit with the person doing the task and watch it end to end, once, without interrupting. It's remarkable how many steps appear in no manual anywhere and only surface by looking.

Then we measure: how many times a month, how long it takes, how many people touch it and what happens when something goes wrong. That last point is what decides whether a project takes two weeks or two months.

How we implement

The part that removes the most hours goes first.

We start with the stretch that has the highest volume and the clearest rules, and get it running in production before touching the next one. The team sees results in weeks rather than quarters, and we get to validate the rules against real data instead of assumptions.

Everything we automate keeps a log and raises alerts. An automated process that fails silently is worse than a manual one, because nobody finds out until the problem is large.

Before / after

What comes back, in hours.

These are the four processes we get asked about most, with the time they take before and what's left after. They're here to calibrate the order of magnitude: if your process resembles one of them, the figure won't be far off.

The “after” is time the team gets back for something else, not one fewer person on the payroll.

Before

Chasing quotes by hand

8H / month

After

Automatic

0H

Before

Preparing the monthly report

6H

After

Generated automatically

12M

Before

Processing requests

3H / day

After

Automated

15M

Before

Copying orders into the ERP

2H

After

No manual step

0H

Illustrative examples of processes that can be automated, not results from real clients. Once we have client work we can publish, it will appear with names and figures attached.

Afterwards

What changes once it's done.

  • The process stops depending on someone remembering.
  • Transcription errors disappear, because nobody transcribes.
  • Response time goes from hours to minutes.
  • Holidays and sick leave stop being an operational problem.
  • There's a record of what happened and when, which didn't exist before.

What doesn't change: you still need someone to decide on the exceptions. The point is that exceptions become the only thing they handle.

How many hours is your process worth?

The audit works it out in under a minute and shows you how it got there.