Automation
Automation and AI for the daily work
The repeating jobs of the week — orders, email, reports, files — done on their own, and done right.
What is included
- The repeating tasks written down, and an honest view of which are worth automating.
- Scheduled jobs: reports, exports, reminders, clean-ups.
- Incoming email and documents processed — read, classified, filed.
- An assistant answering questions over your own documents and data.
- Things that go out to your customer: confirmations, notifications, PDFs, messages.
- A person in the loop wherever a mistake costs: the machine proposes, the person approves.
- Every run logged, so what happened, when, and with what result is visible.
Where work is lost today
Every Monday somebody opens five tabs, downloads three files, merges them into a spreadsheet and sends a report. Two hours, every week, from a person who could be doing something a machine cannot.
What we build
We find the jobs that repeat in a predictable way and write them so they run themselves at a time you choose. Where the job needs judgement rather than a rule — reading an email, classifying a request, answering a question — a language model does that part, with a person checking anything that has consequences.
Works together with
System integrations
The ERP, the shop, the till and the accountant all saying the same numbers, without anyone retyping them.
Read more
Applications & admin panels
The tool your work actually runs on, which today lives in a spreadsheet only one person understands.
Read more
Custom software
If what you need is on nobody’s list, it is on this page. We write it from scratch, for the way you actually work.
Read moreFrequently asked
Will AI replace my staff?
In practice no, and we do not sell it that way. What goes away is the copying from one window into another. Anything requiring a decision, accountability or a relationship with a customer stays with people — and usually gets done better once they stop retyping numbers.
Where does my data go if a model is used?
You decide, and we write it into the contract. There are three routes: a provider contractually barred from training on your data, a provider inside the EU, or a model running on your own machine. The third costs more and is sometimes the only acceptable one.
What happens when the model gets it wrong?
We assume it will and design for it. Anything with a consequence — money, a legal commitment, a message to a customer — goes through human approval. Anything without one runs on its own and is logged so it can be spot-checked.
Contact
What do you do by hand every week?
Describe one repeating job and how much time it eats.