Which jobs will AI replace in Spain? What the data says (and what it doesn't)

What the data says

Worldwide: one in four jobs exposed, very few replaceable

The most cited study on the question is by the International Labour Organization and Poland's NASK institute, from May 2025. Its conclusions:

• Around 25% of global employment is in occupations exposed to generative AI; in high-income countries, 34%.
• Clerical and office occupations are the most exposed, because much of their work consists of processing information.
• But in high-income countries only 5.5% of employment is in occupations with high potential for full automation. For the vast majority, the expected outcome is that the job is transformed, not eliminated.

In Spain: more exposure, but no mass job losses (for now)

The quarterly labour market observatory by FEDEA and BBVA Research, published in September 2026, gives the most recent picture:

• Occupations most exposed to AI account for 30% of employment in Spain.
• Those occupations have grown faster than total employment since 2022.
• There is no mass destruction of jobs. Where there are signs is entry into the labour market: junior programmer vacancies have fallen by around 33%.
• The authors themselves describe it as a first snapshot of a process that is still incipient.

Companies are already hiring with AI in mind

According to Indeed data, 13.5% of job ads published in Spain mention artificial intelligence, more than double the figure in the United States (6.2%). And Microsoft's Work Trend Index 2026 puts at 13% the Spanish AI users who already integrate agents into complete work processes.

AI replaces tasks, not jobs

Look at each job as a list of tasks and the pattern is clear. What AI handles well is what repeats, follows rules and works with information:

• Copying data from one system to another.
• Sorting email and deciding who should handle it.
• Checking that documents are complete.
• Answering clients' most common questions.
• Preparing drafts, summaries and reports from existing data.

What stays human is what requires judgement, relationships and accountability: deciding, negotiating, handling odd cases, talking to an angry client and answering for what gets signed.

An administrator who spends half the day copying data and chasing documents doesn't disappear when those tasks are automated: they spend that time on what they never had time for before.

What it means for an SMB

For a small or mid-sized company, the useful question isn't "which jobs will AI eliminate?" but "which of my team's tasks could a system do?". That's where the digital employee comes in: an AI agent with a defined role that takes on the repetitive tasks of several people, inside their tools and with human approval on anything important.

The usual effect isn't a smaller team but one that grows less than the workload does and spends more time on what creates value.

How to prepare

1. Take stock of tasks, not jobs. For a week, have each person note what they repeat and how long it takes. We explain how in where to start with process automation in an SMB.
2. Put numbers on it. The cost of repetitive work is usually surprising; you can calculate it with the savings calculator.
3. Automate first what takes the most hours and needs the least judgement.
4. Train the team. Since February 2025, the EU AI Act requires staff using these systems to be sufficiently trained.
5. Redesign roles towards work that needs judgement and client contact, and look after the training of people starting out: they are the first to feel the change.

AI won't decide on its own which jobs survive. Companies will, depending on how they use it.