Bots and automation where the standard ends

Where the standard ends

Every piece of software has a limit, and beyond it comes “that cannot be done”. This is where our part of the work starts: we write the script, the bot or the integration that does exactly the one thing your process trips over. It is the reason clients stay with us after the implementation.

A bot makes sense when a task comes back every month and eats people's hours. For something that happens twice a year, we'll advise against it ourselves.

What we automate most often

• Exceptions after capture. A supplier whose invoice layout breaks recognition; the bot corrects those in batches instead of a person going through them one by one.
• Changes across hundreds of documents at once: recategorising, changing a dimension, swapping a scheme after the chart of accounts changes.
• Reports at an agreed hour. “What is missing for August”, “what has been waiting for approval longer than a week”, “which companies have not sent their bank statement”. By e-mail, without anyone having to ask.
• Moving data between systems that nobody has connected before.
• Consistency checks before the month-end close.

How we price it

First we work out how much time the task takes today. If it comes to less than a few hours a month, we say the bot won't pay for itself and we don't issue a quote. That happens more often than you might expect from a company that makes its living from automation. But we'd rather do three things that pay for themselves than ten that make an impression.

What we say before anything is signed

A bot is only as stable as the system it stands on. When the vendor changes the layout of the data, the solution has to be corrected, and that is part of ongoing support rather than a fault. We say so in advance, so that the first update surprises nobody.