AI and Digitalisation for Employment Support in Belgium and Greece

1. Assessment and recommendations

Abstract

Minimum Income Scheme (MIS) beneficiaries in Belgium and Greece experiencing weaker labour market attachment, limited career progression opportunities, and multiple employment obstacles. Drawing on an assessment of current practices, international experience, and in-depth administrative data analysis, this report examines how digital solutions enhanced by Artificial Intelligence (AI) can help address these challenges by improving job matching, strengthening career guidance, and supporting more sustainable pathways into employment. This report proposes concepts for digital job matching and career recommendation tools for Belgium and Greece that provide effective and data-driven support to MIS beneficiaries, while ensuring that these tools are well integrated into the governance structures and institutional frameworks for employment and social support.

1.1. Belgium and Greece demonstrate strong foundations for digital job matching and career progression tools for MIS beneficiaries

1.1.1. Belgium and Greece have well-established systems to support low-income people and households

As many other EU and OECD countries, Belgium and Greece have established MIS to support people living in low-income households, with the objective of reducing poverty and social exclusion. In both countries, these schemes affect a sizeable share of their populations. In Belgium, the number of MIS beneficiaries more than doubled between 2003 and 2023, and around 1.6% of the population depended on the scheme at the start of 2024. In Greece, around 240 000 individuals in 172 000 households received the benefit at the end of 2024.

The MIS in Belgium is designed at the national level but responsibility for benefit administration and service delivery lies with the public centres for social welfare and varies across municipalities, reflecting differences in local capacity, resources, and practices. The scheme aims to ensure a minimum level of income for eligible low-income households through financial support, while also promoting social integration through social inclusion and employment support. Employment support is delivered by the regional public employment services (PES) – VDAB in Flanders, Actiris in the Brussels-Capital Region, Forem in Wallonia, and ADG in the German-speaking Community. Employable beneficiaries can also be placed in public works schemes by the local public centres for social welfare provide beneficiaries with temporary subsidised jobs (referred to as Article 60 and Article 61 measures). These schemes are intended to promote sustainable labour market inclusion, including through the accumulation of social security rights that allow beneficiaries to access PES support.

Greece operates a well-established MIS within a centrally governed framework, where the policy design and scheme architecture are set at the national level, while implementation is supported by a network of local Community Centres, which serve as primary access points for beneficiaries. Similarly to Belgium, the scheme is structured around three core pillars – income support, social inclusion services, and Active Labour Market Policies (ALMPs). ALMPs are delivered by the Greek Public Employment Service (DYPA) through its local Employment Promotion Centres. Support for employment integration by DYPA is standardised for all registered jobseekers, regardless of their MIS status. It includes a profiling phase, a meeting with a job counsellor, the development of an individual action plan, and referrals to appropriate services. Additional targeted support can be provided for jobseekers belonging to special social groups, including via a small number of dedicated offices.

1.1.2. Belgium and Greece operate digital and AI-enabled tools for employment support that are not tailored to the needs of MIS beneficiaries

Despite differences in their level of digital maturity, Belgium and Greece have both made notable progress in modernising their IT infrastructure, particularly since the COVID‑19 pandemic.

Belgium has a highly advanced digital infrastructure for social protection and labour market systems. Regional PES have developed sophisticated AI-driven tools for profiling, skills identification, job matching, and career guidance. Advanced tools also help employers create and manage job vacancies. In addition, Belgium leverages AI for monitoring labour market trends and generating insights to support policy and decision making. However, a key challenge for supporting MIS beneficiaries lies in the fragmented institutional landscape. Local public centres for social welfare operate independently and rely on different systems and practices, while employment services are delivered by separate regional PES. As a result, while the existing digital infrastructure supports many employment and social protection functions effectively, it is less well equipped to support co‑ordinated service delivery for MIS beneficiaries, particularly the ones who are not part of the social security system.

The MIS in Greece is underpinned by robust digital infrastructure, particularly through an online platform that enables automated eligibility checks, continuous monitoring, and fraud prevention. However, this system remains largely limited to administrative functions and has not yet been leveraged for identifying needs for support services or job matching. Greece also benefits from a range of PES digital tools for jobseekers including for registration, profiling, counselling as well as emerging job-matching applications. However, these solutions are not necessarily well connected with each other and are not designed or tailored to the specific needs of MIS beneficiaries. As a result, despite the presence of multiple digital solutions, there is currently no dedicated, integrated digital pathway to support the job matching and career progression of MIS recipients. At the same time, ongoing modernisation efforts further reinforce the momentum for introducing dedicated data-driven matching and career recommendation solutions, including for those with weaker labour market prospects.

1.1.3. Gaps in data coverage and interoperability may constrain their use in job and career recommendation tools

The performance of job matching and career recommendation tools critically depends on access to timely and quality data. Belgium and Greece possess rich administrative data on MIS beneficiaries, but much of this information is collected for specific administrative purposes rather than job search and career guidance, or data-driven decision making.

Belgium benefits from a mature and highly integrated data infrastructure, co‑ordinated by the Crossroads Bank for Social Security (CBSS), which supports secure data sharing and helps ensure data quality and consistency. Administrative registers contain rich information on employment histories, skills, qualifications, demographics, and job preferences, providing a strong basis for digital job and career recommendation tools. Demographic and household data are particularly robust and could support efforts to identify and mitigate potential biases in automated systems. However, existing datasets provide only limited information on informal and transferable skills, work readiness, and other factors that may be particularly relevant for MIS beneficiaries and can limit the effectiveness of employment support. In addition, some variables are not systematically collected or consistently coded, limiting their use in automated processes. Labour market data provide detailed insights into formal employment trajectories, wages, and employer characteristics, but some datasets are only available with a time lag due to the existing reporting structures. On the demand side, vacancy data remain dispersed across multiple platforms rather than integrated into a single system.

In Greece, recent investments in digitalisation have improved the availability, quality and operational efficiency of administrative data. Administrative registers contain extensive information on the competencies, employment histories, and demographic characteristics of MIS beneficiaries, providing a strong basis for data-driven job and career recommendation tools. Nevertheless, inconsistencies in older records, together with limited information in the available administrative data on subsequent employment in public sector employment and self-employment, constrain the ability to fully track long-term employment trajectories. On the labour demand side, vacancy data are systematically collected by DYPA, although coverage remains limited because not all employers advertise vacancies through DYPA’s platform. Furthermore, the coexistence of multiple systems across Community Centres, municipalities, and DYPA leads to partial interoperability and fragmented information flows, which may hinder more effective labour market integration of this target group. Although the National Interoperability Centre provides infrastructure for automated data exchange through web services, its use remains confined to specific operational processes.

1.2. Understanding MIS beneficiaries’ labour market pathways helps designing effective tools that support their labour market integration

1.2.1. Entering paid employment takes time for MIS beneficiaries

Analysis of administrative data generate valuable insights into MIS beneficiaries’ profiles, their registration patterns with the PES as well as their employment trajectories. The analytical approaches applied in the two countries were driven by the data availability in each of them. In Belgium, the analysis relied on tailored aggregate data provided by the CBSS. In Greece, it was possible for the first time to link microlevel data from the MIS register to employment, unemployment and ALMP data for research purposes.

In Belgium, more than half of new MIS entrants remain on the scheme one year after registration, underlining the persistence of benefit receipt. In terms of trajectories, only around 13% of MIS beneficiaries transition into regular employment within a year, compared with nearly 40% of unemployment benefit recipients, highlighting the greater barriers they face in accessing stable work. Labour market outcomes are systematically stronger in Flanders and among beneficiaries with higher qualifications, while young people, older recipients, migrants and beneficiaries in Brussels and Wallonia fare less well.

MIS beneficiaries who transition into employment in Belgium often enter lower-quality and lower-skilled jobs concentrated in the service economy, particularly in administrative and support services, trade, hospitality, health, education, and social care. Employment in goods-producing sectors and higher-skilled occupations remains limited, reflecting both labour market barriers and skill constraints. Compared to unemployment benefit recipients, MIS beneficiaries are more often concentrated in manual service occupations rather than white‑collar roles. These patterns point to persistent inequalities not only in access to employment, but also in the quality and stability of available job opportunities.

In Greece, two in three MIS beneficiaries are already registered as unemployed prior to applying to the scheme and are more likely than other jobseekers to be classified by DYPA as facing additional labour market disadvantages. While around one‑quarter of MIS beneficiaries are employed at some point during their MIS spell, their earnings are often insufficient to lift household income above the eligibility threshold, resulting in continued benefit receipt.

MIS beneficiaries transition into employment at substantially lower rates than other registered jobseekers. While around seven in ten jobseekers registered with DYPA find work within two years of registration, only three in ten MIS beneficiaries do so. There is also a mismatch between the occupations beneficiaries seek and the jobs they eventually secure, with many entering lower-skilled and lower-paid roles such as cleaning, helper, and personal service occupations. In terms of employment conditions and compared to other jobseekers, including groups with severe employment obstacles, MIS beneficiaries are more likely to work part-time, in shifts, and in sectors such as hospitality and in micro and small enterprises. While these jobs provide labour market experience and a source of income, ensuring access to stable and quality employment remains a key challenge.

1.2.2. MIS beneficiaries do not experience employment stability or career progression

The analysis of labour market trajectories in both Belgium and Greece indicates that MIS beneficiaries follow diverse pathways over time, with some transitioning into employment while others remain detached from the labour market. Even among those who find work, employment trajectories are often unstable and progression towards sustainable, quality jobs remain limited, reflecting the persistent barriers many beneficiaries face.

In Belgium, while many MIS beneficiaries eventually exit the scheme, one in three continues to rely on the scheme several years after entry. Transitions into employment may vary by region, educational attainment, and migration background. Three years after entry, former MIS beneficiaries remain concentrated in manual occupations and administrative and support services, sectors characterised by low formal entry requirements, with limited progression towards more diversified and higher-quality employment. Although, the Article 60 programme plays a key role to enable the most disadvantaged MIS recipients to transition into the unemployment insurance system, many programme participants continue to experience weaker employment outcomes and limited progression opportunities. Three years after entering unemployment benefits, only around one in three former Article 60 participants are in employment, compared with one in two beneficiaries who did not participate in the programme. Former participants are also much more concentrated in manual occupations, accounting for around 70% of employment outcomes compared with 45% among non-participants. These findings suggest that many MIS beneficiaries face substantial and persistent barriers to sustainable labour market integration even after entering employment.

In Greece, most transitions into employment occur through registered unemployment, a pattern that likely reflects both the role of DYPA and the influence of registration requirements for benefit eligibility. However, similarly to Belgium, entry to employment does not guarantee lasting labour market integration. Only 38% remain continuously employed one year after entering work, while more than half experience at least one occupational change within the three years following their first post-MIS job. Although some improvements in job quality are observed over time, earnings growth remains modest, particularly among older beneficiaries. Participation in ALMPs also remains limited, with fewer than one in ten jobseekers taking part in at least one programme (excluding training). This suggests that registration with DYPA does not, in most cases, translate into participation in ALMPs.

1.2.3. There is a need for digital tools to recommend career pathways and support services in addition to referring users to specific vacancies

In the case of Belgium, such job and career recommendation tools for MIS beneficiaries should support early intervention by providing timely information on vacancies, training opportunities, and career pathways, before reliance on the scheme becomes long lasting. To improve employment outcomes, these digital tools should provide tailored recommendations based on individual skills profiles and regional labour demand, while integrating pathways to upskilling and reskilling opportunities. Given the concentration of MIS beneficiaries in lower-quality service jobs, these tools should not only facilitate access to immediate vacancies but also highlight opportunities in underutilised sectors and support progression towards more stable and higher-quality employment. Integration with training offers and certification recognition systems could be particularly valuable in helping beneficiaries transition into more sustainable and non-manual occupations.

Similarly, in Greece, job matching and career recommendation tools could support MIS beneficiaries in identifying additional employment opportunities to complement their existing jobs, exploring pathways towards more stable and better-quality jobs, and recognising transferable skills that enable mobility across occupations. Given that some transitions may require upskilling or reskilling or additional experience, digital tools should also facilitate access to relevant training and ALMPs. At the same time, the design of these tools should account for the low digital skills and accessibility barriers often faced by MIS beneficiaries.

1.2.4. Other EU and OECD PES provide inspiration for tailored job and career recommendation tools

PES across EU and OECD countries increasingly rely on digital tools enhanced by AI to support both the immediate goal of matching jobseekers to available vacancies and the longer-term task of guiding them towards sustainable career development. AI-based job matching tools are relatively mature, while career recommendation tools remain at an earlier stage with significant growth potential.

Many PES including those in Austria, Finland, France, Korea, the Netherlands and Sweden, use digital and AI-enhanced tools to support jobseeker profiling, skills identification, job search, competency-based matching, vacancy management, information provision and personalised career guidance. These tools combine structured and unstructured data to identify relevant opportunities, highlight transferable skills, and support counsellors in identifying suitable jobs. These countries have also introduced career recommendation tools that help jobseekers explore adjacent occupations, identify training needs, and plan pathways towards more sustainable employment. Such functionalities are particularly relevant for MIS beneficiaries, who often face complex barriers and fragmented labour market trajectories.

International experience also highlights lessons for implementation. Inclusivity and accessibility are at the forefront of the design of these tools. Countries such as Austria, Germany, the Netherlands as well as Belgium itself (Flanders) place particular emphasis on accessibility, transparency, user feedback, and safeguards against algorithmic bias to ensure that digital tools remain trustworthy, and responsive to users’ needs. Similarly to the progress already made in Belgium (Flanders), France and Israel have focussed on improving data quality through the integration of career data and the adoption of common standards such as the European Skills, Competences, Qualifications and Occupations (ESCO), while strengthening IT and analytical capacity, often in partnership with the private sector. Experiences from Korea and Lithuania showcase the importance of embedding digital tools within broader employment and social support services, recommending jobseeker pathways rather than only vacancies. To ensure accessibility to digital tools, many PES have introduced dedicated support for users with low digital skills, including assistance by PES staff and mobile‑friendly interfaces.

1.2.5. This report proposes a concept for a personalised employment and career guidance system

Drawing on the country assessments, international evidence, and data analysis, this report proposes a comprehensive concept for AI-powered digital solutions. The objective is to support MIS beneficiaries and other jobseekers with severe employment obstacles in accessing employment and progressing in their careers, while delivering benefits at both the individual and institutional levels. Rather than creating multiple standalone tools, the concept emphasises a unified platform that brings together all key services through one interface, improving user experience, reducing fragmentation, and ensuring interoperability, scalability, and long-term sustainability.

The proposed platform is built around five core functional pillars: personalised job matching, enhanced vacancy management, identification of jobseekers’ skills and employment barriers, access to labour market insights, and recommendations for career pathways and progression opportunities. By leveraging AI across these functions, the platform can create more complete and structured profiles of jobseekers, improve the quality and visibility of vacancies, support counsellors and social workers in identifying individuals requiring additional support, and generate more precise job matching outcomes. Beyond connecting individuals with current vacancies, the platform can also translate labour market intelligence into personalised career and training recommendations, helping users make informed decisions about sustainable employment and career progression.

The report outlines country-specific implementation approaches by recognising the different institutional and governance structures in Belgium and Greece. In Belgium, where employment services are decentralised, the preferred approach would be to connect existing regional systems through a federally co‑ordinated interoperability layer. A shared governance framework would ensure common standards and interoperability across regions, while allowing regional PES to retain full autonomy over service delivery. It would also promote mobility, thereby benefiting users to explore opportunities beyond their regional borders. In Greece, the platform could either build on the existing DYPA infrastructure or be developed as a new system, depending on technical feasibility. The tool should eventually function as a unified national platform for all users, beyond MIS beneficiaries. A joint governance model is proposed to strengthen links between the MIS, ALMPs, and employment support services.

Where a fully integrated platform is not feasible, the report proposes a standalone AI-powered career recommendation tool as an alternative. This solution would provide personalised career guidance and options beyond immediate placement, training and ALMP recommendations, independently of a broader matching system, while still delivering significant value to users and supporting informed career development and progression.

Key policy recommendations

Develop integrated and user-centred job matching and career recommendations tools

• Prioritise an integrated digital environment in which job matching, career recommendations and other relevant functionalities are connected through the jobseeker’s existing account, building on the tools already developed or under development and ensuring that MIS beneficiaries can benefit from the same functionalities available to other registered jobseekers. Where additional functionalities are needed to address the particular barriers faced by MIS beneficiaries, these should, where feasible, be integrated into the existing digital environment.

• Design and further develop tools around users’ needs, ensuring that recommendations take account of skills including transferable ones, work experience, preferences, constraints, and employment barriers, opportunities with strong current and future labour demand or persistent recruitment difficulties.

• Use digital tools to support, rather than replace, the judgement and expertise of counsellors and social workers, and their personal interaction with jobseekers, embedding AI-generated insights within case management and personalised support.

• Ensure that digital solutions remain accessible and usable, particularly for users with low digital skills or multiple disadvantages, including through assisted use, alternative access channels and targeted user support.

• Ensure that the developed solutions support the full employment pathway of MIS beneficiaries, from the early identification of barriers to employment through to post placement and career progression.

• Integrate job and career recommendation tools with social services, training, ALMPs and certification recognition systems, to help MIS beneficiaries navigate available support and access the different services needed to move into higher-quality and more sustainable employment.

• Complement digital tools with broader measures that improve job quality, strengthen the targeting and effectiveness of ALMPs, and expand access to social support services for MIS beneficiaries.

Establish effective governance and implementation arrangements

• Establish clear governance structures, roles, and responsibilities to ensure that AI-powered job and career recommendation platforms operate securely, transparently, and in line with national (and regional, in the case of Belgium) social and labour market policies.

• Implement new functionalities gradually, starting with a limited set of high-priority features and expanding them based on implementation experience, institutional and technical readiness, available resources and user feedback. Support this with a clear roadmap with defined objectives, timelines, and resource requirements.

• In Belgium, prioritise functionalities that provide personalised vacancy, social support, training and certification options based on MIS beneficiaries’ skills profiles and regional labour demand, with a particular focus on identifying routes into higher-quality occupations and sectors with strong labour demand that are not commonly entered by MIS beneficiaries.

• In Belgium, connect existing regional employment service systems through a federally co‑ordinated interoperability layer, supported by common governance and data standards to facilitate cross-regional mobility and service co‑ordination.

• In Greece, build on existing and emerging tools functionalities, including the soft skills assessment and AI-supported job description tools, to strengthen the identification of transferable skills and skills gaps, recommend relevant social services, training and ALMPs, and improve vacancy management. This could include further integrating these functionalities to broaden access to suitable job opportunities and help employers identify suitable candidates who might otherwise be overlooked.

• In Greece, build on the DYPA’s existing and emerging digital tools to strengthen the integration and interoperability of job matching, career guidance and labour market intelligence functionalities, and establish joint governance arrangements linking the MIS, ALMPs and employment support services.

• Pilot and test the new tools before large‑scale deployment, engaging key stakeholders, such as employer organisations, associations representing underrepresented groups, trade unions, and employment and social service staff, throughout the design and implementation process.

• Provide counsellors and social workers with training on how to interpret and use recommendations generated by the tools, integrate them into case management, understand their limitations and avoid over-reliance on automated recommendations. Provide jobseekers and employers with guidance and ongoing user support, including help lines, online assistance and assisted access for users who require additional support or have low digital skills.

Ensure access to timely, high-quality, and linkable data for the job and career recommendation tools

• Establish a dedicated legal framework for linking and reusing administrative data across institutions to support AI-enabled job and career recommendation tools, in compliance with national and European regulations, including the EU AI Act.

• Define clear rules on data access, data sharing, institutional responsibilities, and safeguards for the operation of AI-enabled employment services and ensure that users are informed about how AI-supported recommendations are generated and how these recommendations are used.

• Prioritise the integration of data on skills, employment histories, ALMP participation, employment barriers, and labour demand, including labour market intelligence, to support personalised recommendations and respond to evolving skills needs.

• Build on existing classification frameworks and data standards, to further promote consistent coding of skills, occupations, and employability factors across relevant systems and data sources, supporting interoperability and data quality.

• Use longitudinal employment and skills data to strengthen career recommendation functionalities and identify progression opportunities.

• Use demographic and outcome data to monitor the performance, accessibility, fairness, and potential biases of AI-enabled tools.

• In Belgium, strengthen interoperability and harmonisation between welfare offices, PES, and other relevant institutions through the CBSS, to enable integrated and near real-time data exchange for job matching and career progression purposes.

• In Greece, expand interoperability and data-sharing arrangements between the MIS registry, DYPA, ERGANI and ALMP data to enable real-time linkage and exchange of labour market, social protection, and welfare data for job matching and career guidance purposes.

Fine‑tune the job matching and career recommendation tools based on evidence

• Monitor the performance of the tools from their introduction, including take‑up by MIS beneficiaries and employers, the quality of recommendations provided, and subsequent employment outcomes, and, where feasible, evaluate their causal impact on labour market outcomes.

• Use monitoring and evaluation results, as well as implementation evidence and user feedback to regularly refine the tools, including by enhancing underlying recommendations algorithms, adjusting functionalities, improving the presentation of results, and addressing any biases or gaps in performance across user groups.