A single figure sums up this week the true state of digital health for SMEs in Spain: only 1 in 4 companies enjoys good digital health. According to the latest report published this week on the Spanish business landscape, 75% of companies lack solid digital foundations — and the nuance is key: Spain passes at adopting Artificial Intelligence but fails at the foundations on which that AI should rest. That gap is not a theoretical problem: it is the main reason why so many AI projects end up in the bin.
The 2026 paradox: plenty of AI noise, little digital foundation
The figure comes with a second signal that confirms it: nearly a third of Spanish SMEs, sole traders and accounting firms admit they find it hard to apply artificial intelligence in a practical and secure way. And they say so just as the digitalisation grant voucher is on the table. In other words: there is will, there is budget, but the data and process infrastructure that turns the money invested into measurable results is missing.
As Marco Laucelli pointed out this week in an interview about implementing AI in companies: “the prior step is to digitalise the information and the management systems”. It is not about choosing the biggest AI model or the most fashionable vendor. It is about answering first an uncomfortable question: is your company really ready for AI to work with your data?
Why 3 in 4 companies fail (and it is not the technology)
When an AI project fails to deliver value, the temptation is to blame the model or the vendor. But the accumulated evidence points in another direction. There is an enormous difference between having data and having a prepared digital base:
- Scattered, ungoverned data: information lives in spreadsheets, emails, legacy ERPs and loose PDFs. AI cannot get started without a unified, quality repository.
- Half-defined processes: automating a broken process only makes the rubbish move faster. We saw this in our analysis “Why applying AI to your broken processes only makes the rubbish move faster”.
- Lack of profiles and support: buying a tool is not enough; you need someone who translates the business problem into a use case and measures the return.
- Confusing AI with a magic fix: AI is a powerful engine, but it needs quality fuel (data) and a chassis in order (processes and organisation).
The conclusion is a single one: AI in 2026 is won in the data, not in the demo. Exactly the angle we have been defending at BAOSS since our analysis on data maturity published barely a week ago.
The digital health check-up for SMEs: 5 signs your company is not ready
Before investing in AI, an honest diagnosis is advisable. These are the five signs that betray a weak digital base and which, if not fixed first, will weigh down any project:
1. Your business information has no owner and no single definition
If every department understands “customer”, “margin” or “stock” differently, any later analysis will be disputable. Data governance is the first foundation.
2. You still move data by hand between systems
Manual copy-and-paste tasks, Excel reconciliation and control notebooks are a clear symptom of decoupled processes. Automating that — not AI yet — is the first lever of value.
3. You do not know the quality of your data
Duplicates, incomplete records, inconsistent formats. If you cannot measure the quality of your data, AI will produce bad decisions at scale.
4. Critical processes are neither documented nor monitored
It is hard to optimise or automate a process that lives in one person’s head or in an obsolete procedure.
5. There are no baseline cost or time metrics
Without a measurable starting point, you will not be able to demonstrate the return on your AI investment. ROI is built from day one, not at the end.
The right order: foundations → data → automation → AI
Most companies skip steps. They try to jump straight to AI and get frustrated. The path that works — and that we apply at BAOSS with Spanish clients across different sectors — follows a clear sequence. With that base it becomes possible to close the productivity gap with AI instead of accumulating pilots that never reach production.
| Phase | What it achieves | Example lever |
|---|---|---|
| 1. Digital foundations | Mapped processes, a healthy management system | Unify ERP/invoicing, remove duplicate spreadsheets |
| 2. Data maturity | Single repository, governance and measurable quality | Data warehouse, quality rules, data catalogue |
| 3. Automation | Eliminate manual, repetitive tasks | Automation of reports and reconciliations |
| 4. AI on a solid base | Prediction, prescription, business agents | An agent that anticipates churn or recommends stock |
Each phase builds on the previous one and delivers its own ROI. You do not need to finish phase 1 to start seeing value, but skipping phases is the fastest route to a costly failure.
Case study: from weak digital health for SMEs to AI that decides
Imagine a logistics distributor with 200 SKUs and three regional warehouses. Its stock information lives in an outdated ERP and in Excel spreadsheets that an administrator reconciles every night. They want to “add AI” to forecast demand.
The mistake would be to buy an off-the-shelf forecasting tool. Instead, the work starts by cleaning up the data repository: unifying it, defining the single product reference and establishing quality rules. Then the reconciliation that consumed two hours a day is automated. Only then, on a clean base, is the demand forecasting model deployed.
Measurable result: two hours a day recovered for the team, fewer stock-outs, and a forecast the sales team starts to genuinely use because it trusts the data. That is the virtuous circle: first your company learns to read its data, then AI teaches it to decide.
Summary: from the digital gap to real return
The good news is that the digital health gap is, literally, an opportunity. Companies that put their data and processes in order today will hold an enormous competitive advantage over the 75% still making AI noise without a base. The framework is simple. And if you are looking for an orderly way to orchestrate all of this, discover what BAOSS is, the operating system for business agents.
| Common pitfall | The approach that works |
|---|---|
| Starting with the AI model | Starting with digital foundations and data |
| Buying the fashionable tool | Defining the use case and measuring it from day one |
| Automating broken processes | Cleaning up the process first, then automating |
| Expecting magic in 30 days | Sequencing phases, each with its own ROI |
It is not about “having AI”. It is about being ready for AI to generate value. And that starts long before the model, in the order of your data and your processes.
Want to know your company’s digital health and which steps to take so that AI works for you? Talk to our consulting team and we will help you build the base that makes the return on artificial intelligence real.
Prefer to talk to us directly? Contact the BAOSS team and tell us about your challenge.