AI and Digitalisation for Employment Support in Belgium and Greece

2. Greece and Belgium are well-positioned to adopt digital solutions for better job matching and career progression

Abstract

Even though Belgium and Greece have well-established systems in place to support Minimum Income Scheme beneficiaries including through complementary services and employment opportunities, many beneficiaries still face barriers to labour market access. In addition, both countries – though to a different extent – lack dedicated digital tools to effectively support MIS beneficiaries and other disadvantaged groups. The present chapter discusses four main topics for each country, taking into account their specific context and conditions. Firstly, the institutional setup of the Minimum Income Scheme and its links to other services. Secondly, the existing digital tools and their limitations in supporting the matching of MIS beneficiaries with suitable job opportunities. Thirdly, promising data that could underpin more advanced digital solutions. Finally, key feasibility considerations vis-a-vis the countries preparedness to develop dedicated digital solutions to further support this group.

2.1. Introduction

Minimum Income Schemes (MIS) intend to support those most in need, including by connecting them with complementary services and employment opportunities. Nevertheless, MIS beneficiaries in countries such as Belgium and Greece still face major barriers to labour market access. In addition to individual-level challenges, both countries – though to a different extent – lack dedicated digital tools to effectively support MIS beneficiaries and other disadvantaged groups. Existing platforms, including regional ones in Belgium, typically offer vacancy searches and matching services, but fail to adequately respond to the specific needs of this target group. What is more, job postings often do not align with the profiles of MIS beneficiaries, leading to a mismatch where potential opportunities and suitable candidates are overlooked on both sides. However, challenges do not pertain to matching inefficiencies alone. The availability, quality, and sustainability of jobs accessible to these groups are often insufficient, which constrains even more their effective labour market integration although this largely reflects broader labour market conditions beyond the direct control of the PES. While this aspect falls beyond the introduction of dedicated digital solutions per se, it remains an important contextual factor shaping outcomes for MIS beneficiaries and other vulnerable populations.

The present chapter is organised into two sections focussing first on Greece and then on Belgium, following the same structure. Each section begins with an overview of the minimum income scheme, the types of support it provides, and its linkages to other services. It then examines existing digital tools, with a particular emphasis on the gaps and limitations that prevent them from effectively supporting the job matching and career progression of MIS beneficiaries and other vulnerable groups. The third part highlights promising data sources in both countries that could support digital solutions for job matching and personalised career guidance. The final part includes key feasibility considerations. In the case of Belgium, the discussion takes into consideration the country’s federal and regional institutional structure.

2.2. Greece is sufficiently digitally advanced, the needs are clear, and advanced matching and career recommendations tools are both relevant and feasible

2.2.1. The institutional setup of the minimum income scheme in Greece integrates policy design at the national level with service delivery at the local level

Piloted in 2014‑2015 and implemented nationwide in 2017, the Greek Minimum Income Scheme (also known as Guaranteed Minimum Income (GMI) scheme and as Ελάχιστο Εγγυημένο Εισόδημα in Greek) was designed as a last-resort safety net, supporting households in extreme poverty.

The institutional framework governing the scheme combines centralised policy design and implementation at the national level. The MIS operates under the overall responsibility of the Ministry of Social Cohesion and Family (MinSCFA) – particularly the General Secretariat for Social Solidarity and Combating Poverty, which designs its policy framework and oversees its national co‑ordination. At the local level, implementation is carried out through a network of 363 Community Centres, which serve as entry points for beneficiaries.

Since the scheme’s nationwide rollout, several reforms have strengthened the link between income support, social services, and labour activation. The integration of the Fund for European Aid to the Most Deprived (FEAD) into the MIS’s eligibility rules and operational processes in 2017 allows beneficiaries to be automatically enrolled in complementary support directly through the MIS platform. Since 2018, all unemployed and work-able household members must register with the Greek Public Employment Service (DYPA) as a prerequisite for maintaining eligibility (Government Gazette Β 2281/15.6.2018). Efforts to link MIS beneficiaries with social support and active labour market policies (ALMPs) began with a pilot in 32 municipalities in 2019 and were rolled out in June 2021, alongside IT adjustments connecting the MIS platform with DYPA’s information system (Government Gazette B 3359/28.7.2021).

The Greek MIS is based on three pillars: a) income support, b) social inclusion services and c) ALMPS (Directorate for the Fight Against Poverty, 2024[1]). The scheme is financed through a combination of national resources and, for certain components, European co-financing. It covers single and multi-person households, as well as homeless people who reside legally and permanently in Greece, provided they meet defined income and property eligibility criteria.

Applications to the scheme are submitted electronically. This can be done either directly by the applicants using their tax credentials or through the competent services of their municipality or the Community Centre of their municipality with support from staff.1^,2 The scheme relies on robust digital infrastructure that automatically retrieves data from various national administrative databases to verify eligibility in real time, while enabling continuous monitoring and reducing fraud and duplication. Once approved, the benefit is paid by the Organisation of Welfare Benefits and Social Solidarity (OPEKA) on a monthly basis. Eligibility must be reassessed every six months. Beneficiaries are required to keep their information up to date and continue meeting activation and school attendance obligations in order to remain eligible for the scheme.3 In addition to eligibility, updated information would be essential to maintain accurate user profiles for job matching purposes.

Social inclusion services include complementary social services, benefits and goods such as free medical care and medications for the uninsured, reduced rates on utility bills, and access to essential goods, including food and basic assistance packages. Staff at the Community Centres play a key role as they advise beneficiaries on the availability of such services and refer them to those. Community Centres also provide advisory support on employment, youth opportunities, family and legal issues, educational support, and local social initiatives. Some Community Centres in mountainous or island municipalities operate mobile units that serve residents in remote areas. In addition, some Community Centres have dedicated branches for Roma and migrants providing specialised services tailored to their specific needs.

The final step towards self-sufficiency are ALMPs which are implemented by DYPA. Such support may include counselling services, career guidance, seminars, referrals to jobs and measures like employment incentives, vocational training programmes, work experience programmes and public works.4 DYPA gathers data on all regstered jobseekers, including MIS beneficiaries. Some municipalities also operate integrated labour offices, offering job-search support to their citizens. To some extent, this creates some duplication of efforts or variation in the support available depending on the location. Due to the fact that all these actors, namely the Community Centres, municipalities and KPAs, deploy separate systems with some sort of interoperability, relevant information about MIS beneficiaries may not be consistently shared. This fragmentation may affect the labour market prospects of the target group, thereby highlighting the need for more integrated digital solutions on this front.

DYPA’s local Employment Promotion Centres (KPAs) are responsible for supporting jobseekers with their integration into the labour market. The standard activation procedure is well structured and the same for all jobseekers, whether they are MIS beneficiaries or not. Following registration with DYPA, a jobseeker completes an online profiling questionnaire. With the exception of jobseekers who are assigned as closest to the labour market and therefore receive an automated digital Individual Action Plan (DIAP), everyone else then meets with a job counsellor – either in person at a local KPA, by telephone, or online via DYPA’s digital counselling platform – and jointly develop a DIAP. In addition, DYPA provides specialised services and ALMPs for unemployed individuals belonging to Special Social Groups (EKO), such as persons with disabilities, young people at social risk, young offenders, and individuals rehabilitated who have completed a rehabilitation programme. In addition to KPAs, DYPA has a small number of Offices for Special Social Groups to specifically support vulnerable groups through in-person interactions with specialised counsellors and other staff. Nevertheless, high counsellor caseloads (approximately 665 clients each) limit the depth and frequency of personalised support.

2.2.2. DYPA deploys various digital tools to support jobseekers, albeit these tools are not well connected with each other and to the needs of MIS beneficiaries

Greece has made notable progress in modernising its IT infrastructure in recent years, particularly since the COVID‑19 pandemic, with substantial investments financed through the Recovery and Resilience Facility further accelerating DYPA’s digital transformation. Several tools support jobseekers, albeit these tools have not been designed for nor exclusively used by MIS beneficiaries. DYPA offers more than 80 e‑services, including online registration, submission of applications for benefits and programmes, renewal of unemployment card, and posting and updates of CV. Since 2018, DYPA has employed a rules-based profiling tool to assess jobseekers’ employability and refer them to the appropriate services. The tool draws on data collected through a mandatory online profiling questionnaire. Based on the jobseekers’ responses to this questionnaire as well as some on/off criteria, jobseekers are assigned to one of five categories that reflect their relative distance from the labour market.5 While the questionnaire includes some questions to capture the employment barriers of jobseekers, comprehensive tools to identify the skills – particularly informal, soft, and transferable ones – and needs of jobseekers are currently lacking. A soft skills assessment tool, developed with financing from the Recovery and Resilience Facility and currently in a pre‑launch phase is expected to help address part of this gap and will be accessible through DYPA’s e‑services. In previous years, the profiling process was not always completed consistently, which could lead to limited or ambiguous results and reduce its usefulness for counsellors in their interactions with jobseekers (OECD, 2026[2]). Since June 2025, however, the profiling questionnaire has been integrated into the registration process and is now completed by all jobseekers, ensuring more systematic coverage.

Digital counselling has expanded through the myDYPAlive platform, introduced in 2020 (European Commission, 2021[3]). Counselling on this platform is delivered by specially trained counsellors from various local employment offices around Greece. While the platform provides counselling for both jobseekers and employers, it has been particularly beneficial for individuals with disabilities and other vulnerable groups facing significant barriers to labour market integration. There are several new features, such as specific counselling services with interpretation for foreigners and people with hearing disabilities. Feedback collected through the platform has shown that the vast majority of users are highly satisfied (OECD, 2025[4]). The IAP is a key part of the counselling and referral process at DYPA, and its digitalisation streamlined significantly both its creation and the tracking of underlying actions. While helpful in easing administrative processes, these tools do not yet link to matching systems.

Beyond administrative use, DYPA has started leveraging data to monitor ALMP provision through a dedicated Monitoring and Evaluation System, marking a first step towards a more evidence‑based approach in the design and implementation of these policies. Moreover, DYPA’s data are fed into the Labour Market Diagnosis Mechanism (MDAAE), a foundational tool for designing data-driven labour policies in Greece. Among its various functions, the MDAAE collects, analyses, and visualises labour market data – such as employment, unemployment, and vacancies – and identifies the professions and skills most in demand across sectors, regions, and municipalities. It has been developed by the National Institute of Labour and Human Resources and is currently administered by the unit of Experts in Employment, Social Insurance, Welfare and Social Affairs (MEKY) of the Ministry of Labour and Social Insurance (Mechanism of Labour Market Diagnosis, 2025[5]). Although MDAAE generates valuable labour market intelligence it is not yet widely available in user-friendly formats for beneficiaries or counsellors and is not integrated into matching or guidance pathways.

Since 2022, DYPA has used an AI chatbot (“Daphne”) based on natural language processing to answer the queries individuals and employers have about PES services and measures. Its functionality is helpful but limited to rule‑based information provision and is not connected to user profiles or matching decisions.6

DYPA is currently in the process of modernising its current operational IT system. This includes the implementation of a new Integrated Information System (the so-called OPS3). In addition, the agency has ideas for future digital service improvements including via the use of AI. Such ideas include for instance the development of an AI-based chatbot to support CV creation, the development of “Gaspar”, an AI-based tool for digitalising records, tools for assessing jobseekers’ skills, along with AI elements to help reduce counsellors’ administrative workload, improve links between administrative registers and the use of up-to-date labour market information in service provision. As part of a recently started EU TSI funded project, the OECD supports DYPA (and the French PES), in their ongoing efforts to modernise their service delivery systems, including through the integration of AI tools in line with the requirements of the EU AI Act and the relevant national legal framework (Laws 5188/2025 and 5321/2026). For DYPA, the project will propose a comprehensive AI strategy along with an action plan detailing the concrete steps needed for its implementation.

Despite the introduction of new solutions, matching processes remain largely manual and dependent on co‑ordination between counsellors. Digital matching for vulnerable groups is almost non-existent. DYPA’s website integrates a job-matching application, currently operating within an internal testing environment. This tool aims to connect registered jobseekers with employers drawing directly on the characteristics, qualifications, specialisation and skills recorded in the jobseekers’ profiles during registration and the development of their IAP, alongside information on available vacancies. The application also includes a user interface for DYPA staff. More recently, DYPA has introduced the so-called JOBmatch mobile app. While the app enables employers and jobseekers to search independently, it is not connected to DYPA’s or other administrative registers, does not incorporate beneficiary profiles, doesn’t allow mediation from DYPA’s counsellors and is limited to specific sectors.7 As a consequence, vacancy and employer information is spread across DYPA’s portal, the tourism-sector matching app, and private platforms, resulting in fragmented and incomplete data for matching. To add to this, employer participation, although increasingly improving, is not always consistent, limiting the number of available vacancies. Finally, while career recommendation functionalities are still evolving, DYPA has started to integrate elements of career guidance into its emerging AI-supported tools. These include ESCO-based occupation and skills recommendations as well as an interface with the MDAAE integrated into the Digital AI Assistant for the CV to support more informed career choices.

2.2.3. While administrative data quality is high, fragmentation and limited interoperability constrain systematic sharing and effective use

In Greece, the actors involved in the implementation of the MIS collect information and data within their respective databases for operational purposes. Three main registers collect information related to MIS beneficiaries. First, the MIS platform serves as the primary system for collecting data on MIS applicants to assess their eligibility. It is operated by the Greek e‑Government Center for Social Security (IDIKA) under the supervision of the MINSCFA. Data include details on household composition and size, region of residence, degree of disability, educational attainment, as well as income and assets. Second, DYPA gathers data on all jobseekers – including MIS beneficiaries – who are registered with the PES, supporting DYPA’s operational functions. While the majority of data collected by DYPA – such as unemployment information, data on competences and skills and ALMP participation – are stored in DYPA’s Integrated Information System (IIS), some specific information such as participation in training programmes, is recorded in a separate register managed by the external provider, the Computer Technology Institute and Press (Diofantos). DYPA also store data on job vacancies. Third, the ERGANI employment register, maintained by the Ministry of Labour and Social Insurance, records all employment contracts in the private sector, including those involving MIS beneficiaries, containing information on employment histories and employers. Data exchange between different registers is facilitated by the web services provided by the Interoperability Center (KED) of the Ministry of Digital Governance.8

These systems, however, largely operate in isolation, with data sharing occurring only on an ad hoc basis or when initiated by individual case workers. While KED web services provide some interoperability, their use is still limited to specific processes, such as deregistering jobseekers from DYPA upon entering employment. As far as the data itself is concerned, recent investments in the digital backbone of the Greek public sector have contributed to making it more readily available in digital form, reducing errors, improving efficiency and enhancing quality. However, data quality and consistency for earlier years vary, limiting the possibility to track individual career trajectories over a longer period. On the labour demand side, the coverage of vacancy data collected by DYPA is limited, as only a small proportion of employers currently register vacancies with DYPA, which limits the possibility to match jobseekers with suitable job opportunities. Due to limited data exchange, some information is collected multiple times by different institutions through different means. For instance, both DYPA and ERGANI collect data on individual employment histories, DYPA through information provided by jobseekers, and ERGANI via employer-submitted hiring and layoff declarations. While employment data from ERGANI serve as an important building block for any digital tool designed to improve the matching of MIS beneficiaries with labour market demand, the lack of information on public servants and self-employment limits the ability of ERGANI to fully capture individual employment trajectories.

2.3. Belgium can make greater use of matching and career tools, provided interoperability and multi-level governance challenges are addressed

2.3.1. The minimum income scheme in Belgium is designed at the national level, but its implementation is highly decentralised

Belgium’s MIS (Revenu d’intégration sociale – RIS in French or Leefloon in Dutch) established by law in 2002, is designed at the national level but its implementation is highly decentralised. At the national level, the Federal Public Planning Service for Social Integration (Service Public de Programmation Intégration Sociale – SPP IS) is responsible for regulatory oversight, policy co‑ordination, and financial support to implementing bodies. However, the actual delivery of the scheme takes place locally. Each of the 581 municipalities in Belgium operates a Public Centre for Social Welfare (Centre Public d’Action Sociale – CPAS in French / Openbaar Centrum voor Maatschappelijk Welzijn – OCMW in in Dutch), which administers applications, assesses eligibility and provides both cash benefits and complementary in-kind assistance.

This structure means that policy is uniform across the country, but day-to-day implementation varies from one municipality to another, reflecting differences in local capacity, resources, and practices. Social workers at CPAS/OCMWs make case‑by-case assessments, co‑ordinate with regional employment services, and determine the content of individualised integration plans. In this sense, decentralisation refers to the delegation of operational, case‑management, and service‑delivery responsibilities to the municipal level, even though the legal framework and financing are established nationally.

The scheme guarantees a minimum standard of living for individuals and households without adequate resources, while promoting social integration through inclusion measures and employment support.9 The scheme combines financial assistance with social activation measures, providing immediate income support on the one hand while promoting beneficiaries’ longer-term social and economic integration on the other. Beneficiaries may be required to sign an Individualised Project for Social Integration (Projet Individualisé d’Intégration Sociale – PIIS in French, Geïntegreerd project voor Maatschappelijke Integratie – GPMI in Dutch), outlining the necessary steps towards social and labour market integration.

Individuals can submit their applications online (only for first-time applicants) or at their local CPAS/OCMW. They can either visit the CPAS/OCMW on their own initiative or be referred by another institution (for example, employment or housing services). During the first contact, a social worker in the CPAS/OCMW conducts a preliminary assessment of the applicant’s situation. Then the CPAS/OCMW conducts a social investigation – often including a home visit – to verify the living situation and needs of the applicant. A report is then compiled proposing a decision to the CPAS Council, which formally grants or denies the benefit.10 CPAS/OCMW collect rich information about beneficiaries’ needs and circumstances, but this information is not systematically captured or linked across municipalities or with employment services.

Once enrolled in the scheme, beneficiaries gain access to a wide range of services co‑ordinated by CPAS/OCMW, including housing support, healthcare access, psychological or family support, debt counselling, and ALMPs delivered by the competent regional employment services – namely the Walloon Public Service for Employment and Vocational Training (Le Forem), the Public Employment Service of the Brussels-Capital Region (Actiris), the Flemish Employment and Vocational Training Service (VDAB) or the Arbeitsamt der Deutschsprachigen Gemeinschaft (ADG) serving the German-speaking Community of Belgium. Other services may also include education and training such as language courses, vocational training, or second-chance education.

For beneficiaries who are able and willing to work, CPAS/OCMWs may also place them in Article 60 or Article 61 public-works schemes. In this arrangement, the local CPAS/OCMW acts as the formal employer, providing the beneficiary with a temporary, subsidised job within the municipality, other public service or with an external – usually non-profit – organisation. The duration of these contracts corresponds to the number of working days required to qualify for unemployment benefits and varies depending on the beneficiary’s age (from 312 to 624 days). During this period, the CPAS/OCMW pays both wages and social security contributions. Once this period ends, responsibility for labour market support typically shifts to the regional PES. While the schemes were developed as a way to help beneficiaries gain work experience and re‑enter the labour market, in practice, they often serve as an administrative pathway out of social assistance rather than a stepping stone to sustainable labour market integration. However, this information is not consistently kept and interoperable across municipalities.

2.3.2. Belgium has a wide range of digital and AI-enabled tools, but these systems have not been designed for MIS beneficiaries and remain fragmented across regions

Belgium benefits from a strong digital and data infrastructure, but the way information is organised across federal, regional and local levels shapes what is possible for MIS beneficiaries. Belgium’s social protection and labour market systems operate within a highly developed digital environment, supported by the Crossroads Bank for Social Security (BCSS/KSZ), which enables secure data exchange between autonomous administrative registers. This network allows employment services, CPAS/OCMW and federal institutions to verify information such as income, assets or residency in real time and to co‑ordinate across systems. In practice, however, this infrastructure functions mainly as a backbone for administrative processes rather than as an integrated platform for activation. The decentralised nature of CPAS/OCMW, each with its own systems and practices, means that essential information about MIS beneficiaries’ skills, barriers or employment history (among which professional experience under Articles 60 and 61) is either not recorded or recorded inconsistently, limiting the depth and comparability of digital profiles.

The integration of MIS beneficiaries into employment support has been progressively strengthened in recent years. Already in 2014, a federal circular instructed CPAS/OCMWs to register recipients of the minimum income allowance with the regional PES, except where exemptions apply on health or equity grounds (POD Maatschappelijke Integratie, 2014[6]). More recent reforms have sought to strengthen the implementation of this principle. In Flanders, in particular, a 2024 Decree has contributed to an increase in the number of OCMW clients registered with VDAB (Flemish Government, 2024[7]). The federal unemployment reform that took effect in 2026, including the limitation of the duration of unemployment benefits, provides further impetus for closer integration between social assistance and employment support and for strengthened co‑ordination between CPAS/OCMWs and the regional PES. Against this backdrop, social assistance beneficiaries considered able to participate in the labour market are increasingly expected to be registered as jobseekers with the regional PES, allowing their profile and employment-support needs to be recorded in a structured digital file. The practical arrangements underpinning this co‑ordination vary across regions. In Wallonia, for example, systematic information exchange between the PES (Forem) and the CPAS, supported by information available through the BCSS and a centralised partner space within Forem, enables the specific circumstances of MIS recipients registered as jobseekers to be taken into account and facilitates more co‑ordinated employment and social support pathways.

Regional PES have developed sophisticated AI-enabled tools, but these remain largely confined to the PES environment. Across the three main regional PES several advanced profiling, matching and career-guidance tools are already in use to support jobseekers. VDAB pioneered AI-driven profiling with its “Next Steps” model and later the Kans-op-werk tool, which predicted the risk of long-term unemployment using administrative and user interaction patterns. The use of behavioural data was eventually discontinued due to concerns about fairness and potential biases. Le Forem adopted a similar machine‑learning profiling model in 2022, using personal characteristics and past engagement with employment services to estimate a jobseeker’s likelihood of finding work within six months. While these tools are able to predict accurately the risk of long-term unemployment and support caseworkers in identifying individuals who may need more intensive guidance (Desiere and Struyven, 2020[8]), evaluations highlight certain limitations, including model bias and the challenge of interpreting scores without considering the judgement of the caseworker (Brioscú et al., 2024[9]).

Beyond profiling, regional PES also offer a wide range of AI-assisted tools that help jobseekers understand their skills and navigate labour market opportunities. These include VDAB’s Jobbereik (Job Reach), which uses deep learning and graph analytics to identify how jobseekers’ competencies and skills relate to alternative occupations. It doesn’t take into account only their input occupations but also core transferable skills that can be useful in other occupations and industries, hence providing significant insights into job mobility related decisions. The tool also includes a functionality that suggests relevant training and education pathways to close the gap between the jobseeker’s skills and the skillsets of the desired occupations (Brioscú et al., 2024[9]). A similar AI tool, the Competentiecheck (Competency Check) allows citizens to self-assess their competencies vis-à-vis a specific role (Brioscú et al., 2024[9]), while VDAB’s Orient 2.0 uses an AI-based questionnaire to suggest occupations aligned with individual interests and preferences, leveraging the VDAB taxonomy of occupations and skills.11 Outcome monitoring was integrated into their regular processes to ensure fairness and mitigate potential risks (Superlinear, 2025[10]).

Sophisticated matching tools enriched with AI elements are also in place. Systems such as Jobnet and Talent API used by VDAB can help improve the accuracy of job recommendations. Jobnet is a self-learning algorithm for semantic matching between jobseekers and vacancies, while Talent API, a more advanced AI tool which combines deep learning and natural language processing to generate tailored job recommendations. To do so, it relies on jobseeker information – including work experience, skills and geographic location – as well as vacancy text, including synonyms, to expand potential matches. What is more, the tool factors in the jobseeker’s vacancy-browsing behaviour, in order to identify similar opportunities to the vacancies that jobseekers have already viewed. The tool is considered relatively transparent (compared to its predecessor) as the matching scores are visible to the jobseekers, explaining the rationale behind each match (OECD, 2022[11]).

Sometimes vacancy descriptions can be inconsistent or incomplete. To enhance vacancy data quality, VDAB developed Competentiezoeker (Competency Extractor) using AI. In principle, the tool analyses information entered by users or extracted from their CVs to identify the jobseeker’s skills based on their work experience. An additional feature automatically identifies skills from job postings including the ones that are not explicitly mentioned or are omitted. What is more, its Beroepzoeker (Occupation Finder) ensures that vacancies are labelled with the correct occupation classification. More specifically, the tool estimates the proximity of each vacancy to the 600 occupations in the VDAB’s occupational taxonomy, relabelling job advertisements with the closest matching occupation. This improves the accuracy of job postings for employers and enhances job searches for jobseekers through greater consistency across listings. Since 2023, Le Forem employs a similar algorithm based on natural language processing to automatically classify vacancies posted by employers on the PES website according to the occupational classification. Finally, VDAB is currently developing Kandidatenbereik, an employer-oriented tool that applies deep learning and graph analytics to map candidates who fully or partially match the profile of a posted vacancy.

Belgium harness AI for monitoring and knowledge generation purposes as well. A prominent example within this domain is VDAB – the PES utilises autoregressive neural network models to both measure and forecast the demand for occupations and the associated skills, using data from job vacancy postings.

Finally, to promote responsible and effective use of AI tools, Belgium places strong emphasis on governance as well as staff engagement. For example, informal information sessions are organised by VDAB to familiarise staff with new technologies and foster open dialogue between caseworkers and IT teams. Beyond training, the Flanders region of Belgium has established an Ethics Board to provide guidance to VDAB on AI use.

Despite these innovations, their benefits do not reach MIS beneficiaries. Most AI-enabled and digital tools are embedded within PES systems, which are designed for registered jobseekers more broadly. MIS beneficiaries typically engage first and most frequently with the local CPAS/OCMW, whose administrative systems vary considerably across municipalities and are not consistently linked to regional PES platforms. This fragmentation makes it difficult to develop a coherent digital profile of a beneficiary and limits automated referrals or personalised guidance. As a result, many MIS beneficiaries have limited access to suitable opportunities and receive little structured digital support, despite the advanced tools available elsewhere in the system, including in the regional PES.

This fragmentation leaves important needs unmet, particularly for jobseekers with complex or non-linear employment histories. From a matching perspective, there is currently no unified tool linking CPAS/OCMW records with PES profiling, matching or career-planning systems. Labour-demand data are generally updated quarterly and housed within regional systems, reducing their usefulness for real-time or individual-level matching. Data on soft skills, employability barriers, or progression are often captured informally or in free‑text notes, making them difficult to use in digital tools. As a result, existing systems cannot adequately identify informal or transferable skills, which are common among MIS beneficiaries. Without these insights, jobseekers risk being channelled into low-quality or short-term jobs that do not match their strengths or long-term aspirations.

The absence of interoperable tools can significantly hinder the efficiency of matching and undermine inclusive participation in the labour market. Profiling frameworks differ across regions, vacancy data are dispersed across separate PES systems, and the use of federal social-integration platforms such as Nova PRIMA/Nova+ is only optional. This means that information on Article 60/61 placements, labour market transitions or training participation is not systematically transmitted across institutions. These inconsistencies reduce the ability to co‑ordinate activation strategies and limit the potential of data-driven guidance. Overall, the fragmentation of systems underscores the need for more harmonised practices, interoperable tools and mechanisms that allow federal, regional and local actors to work with shared information. These would generate important benefits for MIS beneficiaries and other vulnerable groups because it would give counsellors and jobseekers themselves access to a fuller picture of individual skills, needs and employment histories, regardless of where the information is stored. This would make it easier to identify suitable opportunities, design personalised activation pathways and avoid repeated assessments or gaps in support when individuals move between institutions. Employers would also benefit from more complete and consistent candidate information, clearer vacancy classifications and smoother referral routes across regions.

2.3.3. Data infrastructure in Belgium is strong, but fragmented local implementation and regional differences create challenges for data harmonisation and consistent records

Belgium’s data ecosystem relevant to the employment pathways of MIS beneficiaries is built around decentralised administrative registers, interconnected through the BCSS. CPAS/OCMW maintain beneficiary records covering personal and household characteristics, income sources, integration pathways, competences and participation in Article 60 employment contracts. While these systems follow federal policy guidelines, differences in local IT systems and data‑entry practices affect data consistency across municipalities. To support harmonisation, the federal SPP IS introduced the shared Nova PRIMA/Nova+ platform for processing MIS applications, although its adoption remains optional and is more common among smaller municipalities (Primabook, 2025[12]).

Regional PES (VDAB, Forem, Actiris, ADG) collect well-structured data on jobseekers, including skills, qualifications, work experience, language proficiency, job preferences and employability constraints. Nevertheless, differences in registration systems and workflows across regions can present challenges for cross-regional comparisons and co‑ordinated service delivery (OECD, 2023[13]). In parallel, the National Employment Office (Office National de l’Emploi – ONEM / Rijksdienst voor Arbeidsvoorziening – RVA) manages data on unemployment spells and ALMP participation, while ONSS/RSZ oversees key employment registers such as DIMONA and DmfA, providing detailed information on employment contracts and earnings respectively. These data are further integrated by Sigedis, which maintains longitudinal career records across social insurance schemes. For MIS beneficiaries engaged in public work schemes through Article 60 contracts, work experience data is recorded in the PRIMAWeb system, operated by the federal SPP IS.

ONSS/RSZ also maintains comprehensive records on all employers and employment contracts in Belgium. At the same time, regional PES systematically record vacancy and employer data, including information on employer participation in ALMPs, placements under Article 60/61 and partnerships supporting the labour market integration of MIS beneficiaries. Although vacancy databases remain regional, an inter-regional exchange agreement enables vacancies to be shared across regions.

Overall, Belgium benefits from rich and high-quality administrative data, supported by strong digital infrastructure and secure interinstitutional data-sharing mechanisms. However, important limitations remain. The optional adoption of shared municipal systems results in fragmented CPAS/OCMW records, while employability barriers and soft skills are often captured as unstructured free text rather than through standardised classifications. Differences in data collection practices, profiling frameworks and IT systems across regions also create interoperability and harmonisation challenges for nationwide digital solutions. In addition, some information relevant to the employment trajectories of MIS beneficiaries is not systematically shared across institutions, limiting the potential for integrated and data-driven matching and career support tools.

References

[9] Brioscú, A. et al. (2024), “A new dawn for public employment services: Service delivery in the age of artificial intelligence”, OECD Artificial Intelligence Papers, No. 19, OECD Publishing, Paris, https://doi.org/10.1787/5dc3eb8e-en.

[8] Desiere, S. and L. Struyven (2020), “Using Artificial Intelligence to classify Jobseekers: The Accuracy-Equity Trade-off”, Journal of Social Policy, Vol. 50/2, pp. 367-385, https://doi.org/10.1017/s0047279420000203.

[1] Directorate for the Fight Against Poverty (2024), Interconnection of the Guaranteed Minimum Income beneficiaries with activation services (3rd pillar).

[3] European Commission (2021), PES practices database, MyDYPAlive: Tele-counselling services to jobseekers and employers, https://ec.europa.eu/social/main.jsp?catId=1206&langId=en (accessed on 30 October  2024).

[7] Flemish Government (2024), Decreet over de activering van leefloongerechtigden via de verplichte inschrijving bij de Vlaamse Dienst voor Arbeidsbemiddeling en Beroepsopleiding [Decree on the activation of social welfare recipients through mandatory registration with the Flemish...], https://codex.vlaanderen.be/Zoeken/Document.aspx?DID=1039692¶m=inhoud.

[5] Mechanism of Labour Market Diagnosis (2025), The Mechanism, https://mdaae.gr/en/o-michanismos/ (accessed on  January 2025).

[2] OECD (2026), Strengthening Individualised Support for Jobseekers Furthest from the Labour Market in Greece, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/e732645e-en.

[4] OECD (2025), Optimising Processes and Services at Bulgaria’s National Employment Agency, Connecting People with Jobs, OECD Publishing, Paris, https://doi.org/10.1787/4e79e9db-en.

[13] OECD (2023), Unleashing Talent in Brussels, Belgium, OECD Reviews on Local Job Creation, OECD Publishing, Paris, https://doi.org/10.1787/7a495020-en.

[11] OECD (2022), Harnessing digitalisation in Public Employment Services to connect people with jobs, OECD Publishing, Paris, https://doi.org/10.1787/367a51f3-en.

[6] POD Maatschappelijke Integratie (2014), Omzendbrief betreffende de verplichting voor de OCMW’s om hun gerechtigden in te schrijven bij de gewestelijke arbeidsbemiddelingsdienst [Circular concerning the obligation for Public Social Welfare Centres (OCMWs) to register their beneficiaries with...], https://www.mi-is.be/nl/wetgeving/omzendbrief-betreffende-de-verplichting-voor-de-ocmws-om-hun-gerechtigden-te-schrijven-bij.

[12] Primabook (2025), NOVA+ document technique., https://primabook.mi-is.be/sites/default/files/documents/NOVAPlus%20document%20technique.docx.

[10] Superlinear (2025), Superlinear and VDAB reinvent orientation test with Artificial Intelligence, https://superlinear.eu/impact/superlinear-and-vdab-reinvent-orientation-test-with-artificial-intelligence (accessed on 7 October 2025).

Notes

← 1.  Households with hosted members, households with unprotected children, the homeless and single‑parent families submit applications exclusively to the municipalities and Community Centres.

← 2.  Accountants may be privately hired to file applications on behalf of MIS applicants for a fee, but they cannot make changes in the system.

← 3.  From November 2018 onwards, beneficiary households have been required to ensure that all minor children are enrolled in school when applying to MIS to help prevent school dropout (Government Gazette Β 2281/15.6.2018).

← 4.  Beneficiaries can also be referred to formal educational pathways, such as enrolment in evening lower high schools or second-chance schools (Government Gazette B 3359/28.7.2021).

← 5.  DYPA is currently exploring the adoption of a more advanced profiling tool, in line with practices used by other EU and OECD PES. However, the plans for the future design of this tool have not yet been finalised. A recent OECD project with DYPA – designed to strengthen support to vulnerable clients via modern digital tools and supplementary services –, has developed a concept for a new digital tool to identify clients requiring intensive assistance (OECD, 2026[2]).

← 6.  Plans are already in place to develop a successor to Daphne, with the aim being to develop and implement a more sophisticated AI chatbot that will act as a digital employment counsellor. While the exact details are yet to be finalised, the aim is to implement a tool that will provide more personalised and tailored information and engagement with jobseekers on the services, measures and unemployment benefits available to them – including tailored recommendations for suitable training options (Brioscú et al., 2024[9]).

← 7.  The app is currently available covering positions in the tourism and catering sectors, with plans for future expansion to other sectors.

← 8.  A detailed analysis of the main data registers, underlying data architecture, and available datasets and variables related to the MIS scheme in Greece is available in the relevant output published on the project’s webpage (Output 2).

← 9.  Eligibility for the MIS is means-tested and takes into account applicants’ income, assets, household composition, residency status, and willingness to engage in an individualised integration plan. Claimants must first exhaust other social security entitlements and demonstrate readiness to work, unless exempted for health or social reasons.

← 10.  Claimants can submit an appeal should they disagree with the decision. If the CPAS Council’s decision is positive, the beneficiary is asked to sign an individualised social integration plan outlining mutual obligations and the necessary next steps.

← 11.  Within two weeks from launch, the tool attracted 10 000 users, user input time was reduced by 80%, the number of questions was cut in half, and the accuracy of recommendations was significantly improved.