Croatia's 1.69 million workers score a weighted average AI exposure of 4.93/10. Tourism's outsized role in the economy - contributing over 20% of GDP - creates a labour market with a larger-than-EU-average share of personal service workers scoring 2.5/10, which provides some structural buffer. But Zagreb's growing finance and IT sector, combined with a clerical workforce scoring 9.0/10 at the sub-group peak, means white-collar workers face concrete near-term risk. Croatia joined the EU in 2013 and the Schengen zone in 2023, accelerating alignment with Western European AI adoption timelines.
Croatia's occupation data comes from Eurostat lfsa_egai2d 2025, collected by the Croatian Bureau of Statistics (DZS - Drzavni zavod za statistiku) using EU-harmonised Labour Force Survey methodology at 2-digit ISCO-08 sub-major group detail. Croatia is a small economy with a distinctive structure: tourism's dominance inflates the service and sales share, while Zagreb's emergence as a regional technology and financial hub has built a substantial knowledge-worker base that faces meaningful AI exposure.
| Occupation Group | AI Score | Workers | Share |
|---|---|---|---|
| ISCO 41 - General and keyboard clerks | 9.0/10 | 49.1k | 2.9% |
| ISCO 25 - ICT professionals | 8.5/10 | 37.8k | 2.2% |
| ISCO 42 - Customer services clerks | 8.5/10 | 27.5k | 1.6% |
| ISCO 43 - Numerical and material recording clerks | 8.5/10 | 74.7k | 4.4% |
| ISCO 24 - Business and administration professionals | 8.0/10 | 53.5k | 3.2% |
| ISCO 33 - Business and admin. associate professionals | 7.5/10 | 94.4k | 5.6% |
| ISCO 35 - ICT technicians | 7.5/10 | 21.2k | 1.3% |
| ISCO 21 - Science and engineering professionals | 7.0/10 | 55.5k | 3.3% |
| ISCO 26 - Legal, social and cultural professionals | 7.0/10 | 69.6k | 4.1% |
| ISCO 23 - Teaching professionals | 6.5/10 | 95.7k | 5.6% |
Within clerical support, general and keyboard clerks (ISCO 41) score 9.0/10 - the highest single score in Croatia's occupation structure, across 2.9% of the workforce. Customer services clerks (ISCO 42) and numerical and material recording clerks (ISCO 43) both score 8.5/10, with the latter the larger group at 4.4% of employment. These sub-groups are concentrated in Zagreb's banking sector, government agencies, and the shared services arms of multinationals that have established Croatian operations.
Business and administration associate professionals (ISCO 33) score 7.5/10 and represent Croatia's largest single high-exposure group by headcount at 5.6% of the workforce. Zagreb has attracted regional back-office functions from firms including Infobip - a Croatian unicorn that processes mobile messaging for global clients. ICT professionals (ISCO 25) score 8.5/10 across 2.2% of the workforce, and business and administration professionals (ISCO 24) score 8.0/10 across 3.2% of employment. This concentration of knowledge work in Zagreb contrasts sharply with the coastal Adriatic tourism economy and is the primary driver of Croatia's above-average exposure score relative to its income level.
Croatia presents one of the clearest dual-economy AI risk profiles in Europe. Zagreb is a modern Central European capital with a banking sector, technology companies (Infobip, Rimac Automobili, Nanobit), a stock exchange, and a growing startup ecosystem. The workers in this economy - professionals, clerks, IT staff, financial analysts - face the same AI substitution dynamics as workers in Vienna, Prague, or Warsaw.
The Adriatic coast is an entirely different labour market. Split, Dubrovnik, Zadar, and the hundreds of smaller coastal towns run on tourism for 5 to 7 months per year. Hospitality workers, tour guides, boat operators, waiters, accommodation managers, and seasonal cleaning staff make up a large share of the coastal workforce. Personal service workers (ISCO 51) score 2.5/10 across 5.9% of the workforce and are physically present in the delivery of their service - a hotel housekeeper, a boat captain, a waiter in a Dubrovnik restaurant cannot be replaced by current AI capability in any near-term timeframe. Sales workers (ISCO 52) are Croatia's single largest occupation group at 7.3% of the workforce, scoring 5.0/10 - a mid-range score reflecting that AI already handles online sales and product recommendations even as in-store roles persist. Together, service and sales roles structurally pull the weighted average exposure down relative to a comparably sized economy without a large tourism sector.
ICT professionals (ISCO 25) score 8.5/10 across 2.2% of the workforce and have grown significantly as Zagreb has established itself as a regional tech hub - Infobip reached unicorn status in 2021, and the Croatian tech scene has attracted substantial venture investment in recent years. Business and administration professionals (ISCO 24) score 8.0/10 across 3.2% of employment, concentrated in Zagreb's banking, insurance, and corporate services sector. Health professionals (ISCO 22) score 5.0/10 across 2.4% of the workforce, buffered by clinical interaction requirements. Teaching professionals (ISCO 23) score 6.5/10 across 5.6% of the workforce - Croatia's largest single professional-tier group by headcount, reflecting a sizeable public education sector.
Croatia's physical economy - construction, tourism infrastructure, trades, and personal services - provides a substantial buffer for a large share of the workforce at 2.0/10 or below.
| Occupation Group | AI Score | Workers | Share |
|---|---|---|---|
| ISCO 91 - Cleaners and helpers | 1.5/10 | 49.1k | 2.9% |
| ISCO 03 - Armed forces occupations, other ranks | 1.5/10 | 11.5k | 0.7% |
| ISCO 92 - Agricultural, forestry and fishery labourers | 1.5/10 | 16.2k | 1.0% |
| ISCO 94 - Food preparation assistants | 1.5/10 | 4.5k | 0.3% |
| ISCO 96 - Refuse and other elementary workers | 1.5/10 | 11.0k | 0.6% |
| ISCO 53 - Personal care workers | 2.0/10 | 32.6k | 1.9% |
| ISCO 54 - Protective services workers | 2.0/10 | 36.6k | 2.2% |
| ISCO 71 - Building and related trades workers | 2.0/10 | 48.7k | 2.9% |
Cleaners and helpers (ISCO 91) are Croatia's largest single low-exposure group by headcount at 2.9% of the workforce, scoring 1.5/10 - AI plays no meaningful role in this physical work. Building and related trades workers (ISCO 71) score 2.0/10 across 2.9% of employment. Croatia has sustained high construction activity driven by EU infrastructure funds, Adriatic hotel and resort development, and residential demand in Zagreb and Split. Physical bricklaying, concrete work, scaffolding, roofing, and interior fitting are not tasks that current robotics can cost-effectively replicate in Croatia's construction market.
Personal service workers (ISCO 51) score 2.5/10 across 5.9% of the workforce, the tourism-linked group discussed above. Drivers and mobile plant operators (ISCO 83) score 2.5/10 across 5.8% of the workforce - Croatia's role as a transit country between Western Europe and the Western Balkans sustains demand for truck and goods vehicle drivers. Metal and machinery trades workers (ISCO 72) score 3.0/10 across 4.6% of employment and are employed across Croatia's small but diverse manufacturing base - Rimac Automobili's electric hypercar production employs precision engineers and technicians whose manual assembly skills are not near-term automation candidates.
Croatia's 4.93/10 average reflects a genuine structural duality: Zagreb's knowledge economy faces real near-term AI exposure, while the coastal tourism economy operates in a sector where current AI cannot replicate the physical and interpersonal service delivery. Risk velocity of 7.7/10 places Croatia in the "disruption arriving" band - a 3 to 7 year horizon, faster than economies with weaker digital infrastructure but not as immediate as the 1-to-3-year "disruption imminent" tier some higher-exposure EU peers fall into.
If you work in Zagreb's banking, insurance, IT services, or public administration sectors, the risk timeline is real. AI tools for document processing, customer correspondence, data reporting, and routine analysis are being deployed by Croatian banks and multinationals on the same cycle as in Austria and Germany - Croatia's EU membership means these tools arrive within a few years of their Western European rollout, not a decade later. Roles that involve client relationship management, operational judgment, and process ownership are more durable than roles that produce structured outputs from defined inputs.
If you work in tourism, construction, or trades - whether on the coast or inland - the picture is more stable. Recovery resilience of 6.6/10 is a medium level, reflecting EU Cohesion Fund access for retraining and Croatia's fiscal position after over a decade of EU membership, alongside a labour market shortage that pulls some of the same direction as an aging workforce reduces labour supply. The tourism sector's physical service delivery requirement means the roughly 6% of workers in personal services face a much longer AI displacement horizon than their clerical and professional peers. For workers in coastal hospitality, a multi-year window is a conservative estimate - the question of when AI can replace a boat captain or a traditional Croatian restaurant experience is more open than whether it can replace a data entry clerk.
Interactive breakdown of every occupation group, sortable by AI exposure score and worker count.
View Croatia DataYour reaction helps us understand what to cover next.
Thanks for your reaction!
Get new country analyses, data updates, and AI labour market insights. No spam - one email when something worth reading drops.
AI exposure, robotics, offshoring and work-from-home scores are model-estimated: an AI language model (Claude Opus 4.7) scored each ISCO-08 occupation group on 28 May 2026, guided by published research. They are informed judgements, not measurements, and not forecasts of job losses. "Risk velocity" and its time windows come from a digital-infrastructure index (broadband and secure servers), not from occupations or measured AI adoption. Read the full methodology and how the scores were checked.
Spotted an error or know a better source? Report it and we will check it against the source data.