AI automation
AI automation, aimed at the work nobody should still be doing by hand.
Quotes retyped into three systems, invoices checked by eye, a mailbox that is really a queue. This is the service we are newest at, and we would rather say so.
What we mean by automation
Most of what gets sold as AI automation is plumbing. Something happens in one system, a decision gets made, something else happens in another. The interesting part is the decision, and only some decisions need a model.
So the first pass is always the boring one: what actually happens today, how often, and how much of it is a rule rather than a judgement. Rules are cheaper, faster and easier to trust than a model, and a lot of a process turns out to be rules.
Where a model earns its place
The parts that resist rules are the ones with language or mess in them: a supplier email that has to become a structured order, a free-text field five people fill in five ways, a document nobody wants to read. That is where a model does something a script cannot.
It also has to be checkable. An automation that is right most of the time and quietly wrong the rest is worse than the manual step it replaced, so the ones we would build put a person on the exceptions instead of hiding them.
Where we stand on this
We should be straight about this one. We build software, and automation is software plumbing with a model somewhere in it, which is work we can do. What we have not done is publish a client automation project, so there is nothing here for you to read and judge.
If that matters to you, and it reasonably might, the mobile and web pages have work behind them you can look at. Read this page as what we are willing to build, not as a track record.
In this order
What happens, and where the cost sits.
The same three phases whatever we are building. What a first version costs is on the how we work page rather than here.
- 01
Two or three sessions, free
We sit down with you and structure the idea before anyone writes code. Who it is for, what the first version has to do, what it leaves out on purpose. Two or three sessions, and there is no invoice for them. You leave with the idea written down, whether or not we build it.
- 02
A first version
Then we build it. Not a clickable prototype: the design, the mobile or web app, and the backend it runs on, in a state real people can use.
- 03
Then reaching a market
Building the product is the part everyone talks about. Getting it in front of people who pay is the part that decides whether it lives. We have done that inside our own companies, and we are extending it to the products we build for founders. It is new for us, and we would rather say so than sell it as a finished service.
What's included
What you get, whatever the scope.
01
A map of what happens today
The process written down: what happens, how often, and how much of it is a rule rather than a judgement.
02
The plumbing between your systems
The steps that carry work from one tool to the next, inside the tools your process already lives in.
03
A model, only where the rules run out
Language and mess get a model. Rules stay rules, because they are cheaper, faster and easier to trust.
04
Exceptions a person can see
The cases the automation is not sure about, put in front of someone rather than quietly guessed.
Team
Meet the people you'll work with
No account managers in between. The people who design and build it are the people you talk to.
Oscar Fally
Founder · Product & development
Antoine Visart
Design
Elliot Verstraelen
Development
What people ask about automation.
Do we need AI for this?
Often not. If the process is a set of rules, a model adds cost, latency and a way to be wrong. We would rather say that in the first session than build the more impressive version.
Have you done this for a client?
Not one we can point you at. We build software and we have automated our own work. There is no published client automation project on this site, and we are not going to imply there is.
What does it run on?
Whatever your process already lives in, as far as possible. An automation that requires a company to move to new tools first rarely survives contact with the people who have to use it.
Where does the data go?
That is a decision rather than a detail, and one of the first to settle. Which model, hosted where, and what leaves your systems, gets answered before anything is built.
Enough about us, what about you?
Looking for a partner to build your product? Or sitting on a market you know well and cannot build for? Either way, tell us what you are working on.
Let’s talk