Manifesto
The given
word
Ten things we believe about what happens inside a company between the moment someone says yes and the moment someone does it. You don't have to agree with all ten.
-
01
A small company runs on its word, not on paperwork.
In a multinational almost nothing exists until there is a document. In a thirty-person workshop almost everything exists before there is one, and very often there never is. That is not disorder: it is speed, and it is the only competitive advantage an SME has left against someone ten times its size. The problem is not that people talk. It is that afterwards nobody remembers.
-
02
Any software that demands discipline ends up dead.
We have spent twenty years watching tools go in that work beautifully for six weeks. Then a workload peak arrives, people stop filling things in, the data rots, and by March nobody trusts what is inside. A system only survives if it still works on the worst day of the year, and on the worst day of the year nobody is going to fill in anything.
-
03
Whatever isn't captured in the moment is lost for good.
A commitment has a memory half-life of about four hours. After that, either it is written down somewhere or it depends on the recollection crossing someone's mind by chance. That is why capture has to happen where the conversation already is, and not in a tab someone will open later. Later does not exist.
-
04
Priority is an opinion. Consequence is a fact.
Let people set their own priorities and three weeks later everything is a priority and you are back where you started, with more steps in between. What it costs to miss one specific commitment, on the other hand, can almost always be worked out: it is in the tender specification, in the contract, in the hire rate or in the price of an hour of shutdown. Sorting by that is not more sophisticated. It is simply more true.
-
05
Chasing is real work, and we have left it to the people who take the worst of it.
In almost every SME there is one person —usually in admin, usually with none of it written into the job description— whose actual job is to ask again. It is unpleasant work, invisible and critical, and they do it well because they have an excellent memory and no recognition at all. Automating that does not take their job away: it gives them back the part of the job that made sense.
-
06
A tick is not proof.
Marking something as done costs nothing, and that is exactly why it is worth nothing. When the real argument comes —the customer who says it never arrived, the agency that says it has no record— the only thing that holds is the email with its timestamp, the delivery note with its signature or the photo with its date. A commitment system that does not demand evidence is a system of intentions.
-
07
I would rather have a flagged gap than a well-written hallucination.
A language model can fill any hole with something plausible, and that is precisely the function you have to switch off. If no sentence holds up a date, there is no date: there is an open question. A system that says «I don't know this» can be used to decide. One that invents it cannot, even if it is right eighty per cent of the time.
-
08
What your company says must not train anybody's model.
The internal conversations of an SME contain its margins, its purchase prices, its trouble with customers and its arguments. Handing that to someone else's service in exchange for convenience is a decision you take once and cannot undo. That is why the model runs in your own building and the code is open: not so we can say so on a website, but so that you can check.
-
09
Watching commitments is not watching people, and the difference has to be defended.
The same technology that remembers a promise could count how many each person breaks, and that product would be easier to sell to certain directors. We are not going to build it. Partly out of conviction and partly because it would destroy the product: the day the team suspects this exists to name culprits, they will stop forwarding emails and there will be nothing left to analyse.
-
10
Keeping your word is still the best commercial advantage there is.
In a sector where everyone runs late, being the supplier who delivers on Thursday having said Thursday is not a moral virtue: it is a market position. Customers notice it before they notice the price and they remember it for longer. All this engineering —the local model, the clock, the chain of reminders— exists to hold up something rather old: that what is said is as good as done.
We spent thirty years on sites, in workshops and in engineering offices before writing a single line of this product. Everything above we have watched fail in person, and in more than one case we were the ones who failed.
…engine to Innovate
What gets said, gets done.
That is an easy line to put on a website. The hard part is holding it up on a Tuesday in November with three sites running. That is what we built this for.