by Claudine Adeyemi-Adams FIEP | CEO & Founder | Earlybird
INTRODUCTION: THE CASEWORK LEGACY
For decades, employment support has been built on the principle of casework: one participant, one adviser, one plan. This one-to-one model has clear strengths. It enables personalisation, builds trust between adviser and participant, and creates accountability for progress against an agreed action plan.
However, casework is inherently reactive. Structural barriers such as inaccessible childcare, transport limitations, or emerging labour market skills gaps are often recognised late, after they have already hindered many people’s journeys to work.
And because casework consumes significant time and administrative effort, resources are heavily tied to individual-level support, limiting capacity to address systemic challenges.
Across public services, artificial intelligence (AI) is enabling a shift from purely case-by-case intervention to what can be called “systems work”: the ability to spot patterns, anticipate needs, and design proactive responses at scale. This is not a replacement for casework, but an additional lens that can make employment services more effective, equitable, and sustainable.
DEFINING “SYSTEMS WORK”
In this context, systems work means looking beyond the individual file to understand and act on trends across many participants, communities, or programmes. It involves aggregating and analysing data to uncover structural barriers and opportunities and then using those insights to inform policy, programme design, and resource allocation.
The concept is already well established in other domains. For example, in the healthcare sector, population health management uses patient data to predict needs, target interventions, and prevent illness, shifting resources from reactive treatment to proactive care (NHS England, 2023). Or, in New South Wales, Australia, an indigenous-led justice reinvestment initiative, operating regionally in Bourke called Maranguka partnered with the New South Wales Police and other local and federal agencies to securely receive and hold data that ultimately informed the development of community-led initiatives and policies tailored to the specific needs of Bourke, through data working groups. This enabled Maranguka to design programmes successfully targeting the root causes of youth crime (Maranguka Ltd, 2024).
In employability, a systems lens might reveal that:
- Jobseekers in a specific locality with limited public transport access are significantly more likely to drop out of employability programmes.
- Demand for digital literacy support is rising sharply among older workers in a region where all local libraries have closed down.
- Participants with intermediate-level English proficiency are more likely to gain initial employment but struggle with retention after the first month.
These kinds of insights allow interventions to be designed before the next individual experiences the barrier – potentially benefiting dozens, hundreds, or even thousands of people.
HOW AI ENABLES THE SHIFT
While practitioners and managers have always had some awareness of systemic patterns, AI changes the scale, speed, and accuracy with which those patterns can be detected and acted upon.
What could that look like in practice? AI can supercharge data aggregation; bringing together diverse datasets – adviser notes, participant surveys, employer feedback, labour market statistics – all in real time. Manual collation of such data is slow and labour-intensive; AI can make it continuous. You could have machine learning models identifying correlations and anomalies that may not be visible to humans conducting reviews. For example, natural language processing can scan thousands of adviser case notes to highlight recurring themes around mental health, housing instability, or skills gaps.
With predictive capabilities, rather than simply describing what has happened, AI can predict what is likely to happen next. This might include identifying participants at high risk of disengagement, forecasting demand for specific training, or anticipating the impact of labour market shifts. Or imagine simulating scenarios – AI could model “what if” scenarios to test the likely outcomes of policy changes or programme adjustments before they are implemented, helping decision-makers allocate resources more effectively.
IMPLICATIONS FOR SERVICE DESIGN
If we can advance towards systems work in some of the ways set out above, there is potential to completely reshape how employability services are designed and delivered.
Instead of responding when individuals disengage, services could act earlier to support whole groups with targeted measures for example, arranging subsidised transport in an area identified as a dropout hotspot. We could have data-driven insights guiding the deployment of specialist advisers, outreach teams, or employer engagement resources where they will have the most impact.
You could also see commissioners adopting funding models that reward providers not only for individual job outcomes but also for achieving systemic improvements such as reducing a region-wide barrier to employment.
RISKS AND ETHICAL CONSIDERATIONS
Shifting from casework to systems work is not without its challenges.
Focusing on aggregated patterns must not obscure individual needs or reduce the quality of personal support. AI should augment, not replace, the human relationships that underpin successful employment support. Careful consideration would need to be applied to how changes are implemented so that we do not fall into a “one-size-fits-all” approach when a highly personalised service is what’s needed.
It’s also imperative to bear in mind that AI systems are only as good as the data they are trained on. If the data excludes certain groups or embeds historical inequalities, the resulting insights may perpetuate bias rather than challenge it. Building AI systems that solve systemic issues is therefore incredibly difficult (albeit not impossible).
To reduce bias and keep ‘systems work’ equitable, providers can adopt a trust-by-design approach: co-design with affected groups; use representative, regularly refreshed datasets; run pre-deployment Algorithmic Impact Assessments and Data Protection Impact Assessments; and publish Algorithmic Transparency records so stakeholders can see how tools influence decisions. Ongoing monitoring, clear appeals routes, and human-in-the-loop reviews are essential.
Established frameworks already offer fairly concrete steps: UNESCO’s Recommendation on the Ethics of AI emphasises transparency, fairness, and human oversight; the UK Data Ethics Framework guides teams through ethical use across the lifecycle; the UK’s Algorithmic Transparency Recording Standard enables public sector organisations to publish information about the algorithmic tools they are using and why they are using them; and Canada’s federal Algorithmic Impact Assessment operationalises risk scoring and mitigation before systems go live.
The needs and thoughts of participants also have to be considered in this new approach and are too often excluded. Will participants consent to their data contributing to aggregated insights? How can the narrative be built to ensure that there is true transparency and trust built with participants and that they buy into the value created at scale as a result of their data being used to shape future services?
POLICY AND WORKFORCE IMPLICATIONS
The shift to systems work requires not just technology but also changes in policy, workforce roles, and organisational culture. Advisers may increasingly be asked to contribute to service improvement through the insights they gather in daily work, becoming both practitioners and data contributors – this may impact their existing workflows or require upskilling and deeper understanding of the bigger picture. It is likely that commissioners would need new metrics to measure success at the system level, including intermediate outcomes such as barrier reduction or community-level skills growth.
But the greatest potential may lie in linking employability data with other public service datasets – housing, health, education – to address root causes of unemployment more holistically as advocated for by the OECD which repeatedly emphasizes cross-sector data sharing and common data governance as a way to enable more integrated services (OECD, 2019).
Looking internationally, we can see precedents for this shift:
- United Kingdom: Integrated Care Systems use aggregated patient data to identify priority health issues in local populations, reallocating resources to prevention (NHS England, 2022).
- United States: From 2016–2021, 25 county-led pilots in California linked health, housing, justice, and social services data to coordinate support for people with complex needs. By 2018, all pilots had formal cross agency data-sharing agreements, and most used shared electronic platforms combining medical, behavioural health, housing, and service-use information. Some provided real-time alerts (e.g., emergency visits) to trigger rapid, coordinated responses. This enabled agencies to address root causes like housing instability or untreated mental health issues at a population level, rather than relying solely on reactive case management (UCLA Centre for Health Policy Research, 2019).
- Estonia – Public Services: e-Estonia is a national data infrastructure enables agencies to share and act on aggregated insights, making services more joined-up and proactive and X-Road (which is the foundation of e-Estonia) has even enabled cross border data sharing between Estonia and Finland which supports the future development of additional cross-border services.
These examples show that success depends as much on governance, political will, and public trust as it does on technology.
CONCLUSION
Casework remains essential. The empathy, trust, and adaptability of human advisers are irreplaceable. But AI offers the possibility of seeing the forest as well as the trees — identifying patterns that no single adviser or manager could detect, and acting on them to prevent problems before they affect the next participant.
The future of employability services is likely to involve blending casework and systems work – the human-led, one-to one support that meets people where they are combined with the AI-enabled, systems-level insight that shapes a more equitable and effective service for all. This will require investment in data infrastructure, workforce capability, and cross-sector collaboration – but the reward is a service that is both more personal and more powerful.
REFERENCES
Maranguka Ltd, 2024. Maranguka’s Submission: Inquiry into Community Safety in Regional and Rural Communities. Inquiry The Committee on Law and Safety, Parliament of New South Wales. 31 May 2024. justreinvest.org.au, https://www.justreinvest.org.au/wp-content/uploads/2024/06/Marangukas-Submission_-Inquiry-into-Community-Safety-in-Regional-and-Rural-Communities.pdf. Accessed 13 August 2025.
NHS England, 2023. “Population Health Management Guidance.” NHS England, 17 February 2023, https://www.england.nhs.uk/long-read/population-health-management. Accessed 12 August 2025.
UCLA Centre for Health Policy Research, 2019. Health Policy Brief: Whole Person Care Improves Care Coordination for Many Californians. October 2019. UCLA Health Policy Research, https://healthpolicy. ucla.edu/sites/default/files/legacy/Documents/PDF/2019/wholepersoncare-policybrief-sep2019.pdf. Accessed 13 August 2025
ABOUT THE AUTHOR
CLAUDINE ADEYEMI-ADAMS FIEP | Founder & CEO | Earlybird
Claudine Adeyemi-Adams is an impact technology leader and the founder of Earlybird, an AI and Voice technology platform, empowers organisations delivering Government contracts – particularly those supporting people into work – to improve outcomes more efficiently.
Her personal journey from a low-income background to a multi award-winning legal career ignited a deep passion for social change. Having experienced homelessness and navigated the complexities of employment support programmes firsthand, she is uniquely positioned to understand the barriers many individuals face.
Claudine is the IEP Fellow of the Year and Earlybird won Digital Solution of the Year 2024 at the ERSA Employability Awards.