The invisible list

Everything your company has stopped asking of its ERP. And why, from now on, you can start asking again.

Tuesday morning. The head of administration at a sixty-person company opens the ERP, looks at the invoicing screen, and then does what she does every Tuesday: she opens a spreadsheet next to it.

In that spreadsheet are the things the system does not give her. Profitability by project. The alert that an order has been stuck for five days. The three fields she needs on the customer record that nobody has ever added. She has been keeping it by hand for two years and no longer imagines it could work any other way.

That spreadsheet is not a software problem. It is a receipt.

The invoice that appears on no invoice

When a mid-sized company has lived with its ERP for a few years, something appears that nobody talks about, because it has no line in the accounts: the list of improvements it has stopped asking for.

It is not neglect. It is learning. You have asked for things before and you know how it ends: it has to be scoped, you get quoted by the hour, it joins a queue and it arrives late. After three or four rounds of that, any sensible manager does the rational thing — stops asking, builds a spreadsheet on the side, and gets on with the work.

We call that the invisible list. And it is the exact measure of what your system really costs you: not what you paid for the rollout, but everything your company has decided not to do because asking was slow and expensive.

Nobody invoices you for the list. You pay for it all the same, in decisions made with worse information than you could have had. You do not notice it in a month. You notice it in three years.

Why the list exists (and why it is not your provider's fault)

It is worth being fair to the industry here, because nobody's bad faith created the invisible list. A cost structure created it.

A project on an ERP has four stages: understanding what you need, building it, testing that it does not break anything, and leaving it documented. All four consume hours from expensive people, and in a model that sells hours, all four get invoiced. That is why a report costs what it costs. That is why it has to be scoped first. That is why the queue exists.

With that economy, your provider cannot tell you anything else. And with that economy, you cannot do anything other than stop asking.

This is what changed in 2026. And the change is not that artificial intelligence models got cleverer — it is where they are. AI has stopped living in a separate tab, writing emails for a demo, and moved inside the systems where the work actually happens: the ERP, the warehouse, invoicing, accounting.

Of those four stages, the first is still a conversation between people: no machine shortens the work of understanding how you operate. But building, testing and documenting are the three most repetitive and best-documented stages in the software world. And they are exactly the three that supervised agents do in a fraction of the time.

The consequence matters more than the technical detail: the price does not fall because the hour is billed lower, it falls because far fewer hours are spent. It is not the same thing, cheaper. It is the same thing with ten times fewer hours. And once the cost of asking collapses, the invisible list has no reason to exist.

We tried it in our own house

We got here the long way round. We are AiKit Research, an artificial intelligence lab in Tres Cantos, Madrid, and two years ago we had the problem of any fast-growing company: running the business no longer fitted in spreadsheets.

We chose Odoo — not out of romanticism about open source, but because it is modular, it covers everything from CRM to accounting, and it can be extended. That last word is what opened up everything that followed. Because instead of asking a provider for every change, we put our own artificial intelligence systems to work inside the ERP.

Here is what happened, specifically.

Bank reconciliation stopped existing as a task. If you run a fifty-person company, this will sound familiar: reconciling the bank is the most time-consuming activity of the monthly close, ahead of accruals and variance analysis. The industry numbers are uncomfortable — between 20 and 50 hours a month at companies of 51 to 200 employees, four hours per bank account at the market median, and 94 % of finance teams still doing it with a spreadsheet in the middle. In our house, transactions come in on their own, the reconciliation proposal arrives already made, and a person reviews the exceptions. From days spread across the month to one short review.

Payroll and time tracking are a button. They used to be a monthly ritual of preparing, reviewing, sending and correcting. Time records, on top of that, are not a report you can decide not to produce: they have been mandatory in Spain since 2019 and have to be kept for four years. Today the whole process is described inside the system and the mechanical part is executed by the machine.

So far, everything is the same thing, faster. The interesting part starts now.

The CRM fills itself in. Anyone who has run a sales team knows how this story ends: half-finished records, call notes never written up, and on Monday the salesperson reconstructing a whole week from memory. Not out of indiscipline — because nobody has ever sold more by filling in a form. Today our cards are complete and sales activity is visible in real time without the salesperson doing anything. We are not asking them for more rigour: we are taking the admin off their hands. And whoever runs the team has gone from looking at an incomplete picture of last week to looking at what is happening now.

Meetings turned into the company's memory. They are all recorded, transcribed and processed, and out of that come the decisions and the actions, written where anyone can find them. What was decided in a meeting three weeks ago no longer lives in the head of whoever took notes. This module did not come with Odoo: we wrote it ourselves, along with the API layer we use to run the business from outside.

Notice the difference between the first two examples and the last two. The first two save hours: you can work them out on a spreadsheet. The other two save nothing, because they simply were not being done before. They have nothing to compare against. And those are the ones that were on our invisible list.

The acid test

And now the example that measures whether all of the above is hot air.

We had an external accounting firm, a very professional one. We no longer work with them. Not because they did a bad job, but because we run our own accounts inside our own Odoo, with the Spanish fiscal localisation, with artificial intelligence on top and with the controls needed to sleep at night.

I am telling you this because I know exactly how it sounds. Accounting is the last thing a sensible company outsources and the last thing it would dare bring back in-house, because a mistake there is not fixed with an apology. We brought it back. That is the level of confidence we have in what we do — and it is the only honest way to ask you to trust us.

This is how Nodu was born

Odoo is a company's heart. Artificial intelligence is the nervous system it had been missing. We spent two years connecting one to the other in our own house, and the conclusion was hard to ignore: this is not an internal trick from an AI lab. It is what thousands of mid-sized Spanish companies are missing — companies that already have Odoo and an invisible list that keeps growing.

Nodu is doing in your company what we did in ours. It is an initiative of AiKit Research — we are not hiding that, although we are not going to put a powered by on the logo either: we would rather tell it with this story.

And it starts with four rules that sound odd in this industry, all for the same reason: when supervised agents do the repetitive part, the cost is predictable, and a predictable cost allows a predictable price.

The prices are written on the website. A figure, a scope and a date, before you call. Not a bucket of hours and not a "depends on the scope". They are all published.

You start without paying and without anyone touching anything. We look at your Odoo and tell you what you have, what you are not using and what can be solved. If nothing useful comes out of it, we will be the ones to say so. And when we build, we clone your ERP onto our machines and work on the clone: your installation stays exactly as it is, with no permissions and no open ports, with no risk to what already works.

You use it before you commit. We hand it over working and you decide afterwards. The subscription has no lock-in, and it includes usage support — because the most repeated complaint in this industry is not the rollout, it is the silence afterwards.

You buy the Odoo licence from Odoo. We do not resell it and we do not carry a hidden margin inside it. What you pay us for is service.

And something we say just as clearly: what we do not do. We do not touch your accounting history or your opening balances. We do not put our hands into custom development somebody else built for you. We do not sell "unlimited": one active request at a time. Written limits are the only way a low price is believable.

There is one date that is not up to you

We are not going to tell you that doing nothing puts you out of the market. But there is one thing on your invisible list that has stopped being optional: Veri*Factu requires Spanish companies to invoice through a compliant system from 1 January 2027, and everyone else from 1 July. It is the only line on the list you cannot postpone.

If you are going to have to touch your invoicing anyway, it is a good moment to touch it properly.

Getting started costs half an hour and nothing else

The first step is not a sales meeting or a quote. It is a health check of your Odoo: we tell you what is inside, what is missing and what can be solved. At no cost and without anyone touching anything.

After that, thirty minutes with no proposal and no figures — you just tell us how you work and what you cannot see today.

Bring your invisible list. It is the only thing we need to get started. Write to us.