How long before it really works?

The first signals appear between the fourth and the eighth week: specific processes already run by the system, with full traceability. The complete enterprise brain —perceive, coordinate, decide, execute, learn— is built in layers over months. Be wary of anyone promising a transformation in two weeks: what gets delivered that fast is a demo, not an operation.

Do I have to change my CRM, my ERP or my current stack?

No. We connect to what you already have. Migrating the stack is the favourite excuse for delaying a project by three quarters. Your problem is not which tools you use, it is that none of them speaks to the others: the data exists, but nobody reads it together. We build the intelligence layer on top of what you already pay for.

We already have automations. How is this different?

An automation executes a rule you wrote. An enterprise brain interprets context, chooses between alternatives and learns from the outcome. The practical difference: automation breaks when reality changes; the brain adapts and warns you about what you have not seen yet.

Will my team reject it?

Your team rejects tools that add work. Here the system absorbs the administrative part —the copying and pasting, the searching in five places, the reporting— and gives back time for judgement. We start with the process that hurts most internally, not with the one that looks best in a presentation.

What do we need to bring?

Access to the data, an internal owner with the authority to decide, and candour during the assessment. The audit phase is uncomfortable on purpose: it surfaces which processes are undocumented, which data is duplicated and which decisions are taken on instinct. Without that, any AI amplifies the disorder you already have.

Where does my data end up?

On infrastructure under your control and with residency you define. We do not train public models with your information, nor mix it with other clients' data. Every access is logged: who, when, what and why. If a supplier cannot answer this in writing, the conversation should end there.

It is designed in, not bolted on. Data minimisation, a legal basis per process, defined retention and a right to erasure that can actually be executed. We work with INCIBE, Spain's national cybersecurity institute, precisely because security and accountability are verified, not declared.

What happens if the AI gets it wrong?

We assume it will. That is why the operational core is deterministic —rules, limits and hard validations— and the intelligent layer acts inside that perimeter. Decisions with financial or legal impact go through human approval. Everything is audited and reversible: you can see why the system did what it did.

Can anyone extract sensitive information through a conversation?

Access is resolved by identity and role, as in any serious system: the agent cannot show what that user has no right to see. On top of that there is control over which information may leave the perimeter and a full log of every query.

Who owns what gets built?

You do. The logic, the flows and the data belong to your company. No knowledge held hostage and no architecture designed to stop you leaving. If one day you decide to run it without us, you take your brain and its documentation with you.

It works in one department. Will it hold across the company?

Yes, and that is exactly what it is for. At Percent we work with close to 150 professionals, some 200,000 contacts and more than 100,000 business records. The system grows in layers: each new process builds on the data and governance foundation already in place, so the second department costs considerably less than the first.

What return can I expect?

It depends on where the friction is today, and we will tell you if there is none. The return arrives through three routes: administrative hours recovered, opportunities that used to be lost for lack of follow-up, and decisions taken on data instead of on memory. We define the metrics before we start, not when we present the invoice.

What if my business changes?

The system learns, and that is the structural difference. Processes are versioned and the accumulated knowledge is not lost when the process changes. A rigid automation has to be rewritten every year; a brain is re-educated.

Does this replace my team?

It changes what your team does. The AI operates, people supervise and decide. In our own case —Subvia— the company runs entirely on AI with a human team that sets the strategy and validates the direction. It is not fewer people: it is the same people without tasks nobody should be doing.

Why start now and not next financial year?

Because the cost of waiting appears on no budget line. Every quarter without an intelligence layer is another quarter of data generated and not used, and of context lost. Your competitors are not buying tools: they are building accumulated operational judgement, and that cannot be bought at the last minute.

Enterprise FAQ

Questions every
executive should ask

No technical jargon and no brochure promises. The real doubts that come up in a board meeting when building an enterprise brain is on the table.

AI implementation

Most AI projects do not fail because of the technology. They fail because nobody defined which decision was supposed to improve.

Security and governance

An enterprise brain only makes sense if it is a secure brain. Cybersecurity is not an add-on to the project: it is its foundation.

Scale and return

The right question is not how much AI costs. It is how much it costs you every month to keep deciding blind.

Your question is not here?

Let us talk about your case

Every company is fragmented in its own way. Tell us where the information breaks down and we will say plainly whether building an enterprise brain makes sense now.

Get in touch

Ready to apply this to your business?

Eva helps you choose. Our team scopes your project.

Book an audit · €290 ↗

Questions everyexecutive should ask