AI Eligibility Checker

Overview

I worked on a proof of concept with Deloitte to explore how AI could make it easier for people to navigate government services. Focusing on the citizens who face the most friction, the concept brought multiple services into one trusted, conversational entry point that was designed to be transparent, multilingual, and supported by humans when needed.

Organisation

Deloitte

My responsibilities

Problem framing, Journey/System Mapping & Orchestration Logic, Conversational Concepting

Challenge

Citizens often face a fragmented experience; from jumping between departments, to repeating the same information, and getting different answers depending on where they start. With limited data sharing, it can take longer to resolve issues and people can miss out on entitlements. The challenge was to show how responsible AI could reduce this friction by connecting services, supporting staff, and helping citizens get fair, transparent outcomes they can trust.

Research playback insights 1
Research playback insights 1

Approach

Problem framing and principles → Started from real citizen pain points, then overlaid where AI could responsibly support staff and citizens. Set experience principles grounded in trust, inclusion, explainability, and human fallback. Research and ecosystem mapping → With limited user access, triangulated insights from public data, government reports, surveys, and social media discussions. Mapped the Irish public-service ecosystem (Justice, Housing, Education, Social Protection, HSE) to surface duplication, gaps, and handoffs. This highlighted where people need clarity and where fragmented data blocks proactive support. Proto-persona and use cases → Built the concept around a proto-persona, Martha, a newcomer to Ireland navigating visa, healthcare, education, and housing. Used her journey to define priority use cases and show how AI could reduce cognitive and admin load, while keeping fairness, privacy, and human oversight front and centre. Mapping and prototyping → Visualised the AI experience logic through journey maps and system diagrams, defining three progressive layers: Pre-Arrival Guidance (multilingual, trusted info), Personalised Support (tailored guidance across departments), and Proactive Assistance (predictive prompts with human review).

Approach

Problem framing and principles → Started from real citizen pain points, then overlaid where AI could responsibly support staff and citizens. Set experience principles grounded in trust, inclusion, explainability, and human fallback. Research and ecosystem mapping → With limited user access, triangulated insights from public data, government reports, surveys, and social media discussions. Mapped the Irish public-service ecosystem (Justice, Housing, Education, Social Protection, HSE) to surface duplication, gaps, and handoffs. This highlighted where people need clarity and where fragmented data blocks proactive support. Proto-persona and use cases → Built the concept around a proto-persona, Martha, a newcomer to Ireland navigating visa, healthcare, education, and housing. Used her journey to define priority use cases and show how AI could reduce cognitive and admin load, while keeping fairness, privacy, and human oversight front and centre. Mapping and prototyping → Visualised the AI experience logic through journey maps and system diagrams, defining three progressive layers: Pre-Arrival Guidance (multilingual, trusted info), Personalised Support (tailored guidance across departments), and Proactive Assistance (predictive prompts with human review).

Logica UI mock-ups
Logica UI mock-ups

Impact + Outcomes

Approach

The proof of concept translated responsible AI principles into something tangible: end-to-end system journeys and orchestration logic for an agent-driven interface, plus a service vision and prototype for human-centred, AI-enabled public services. It also surfaced practical opportunities for cross-department data integration and governance, which are key enablers for trust at scale. The concept won the Dublin Regional Final of the Here for Good AI Hackathon 2025 and progressed to the national final, sparking new conversations with government clients and reinforcing Deloitte’s position in trustworthy AI design. Personally, it stretched my thinking on AI UX, conversational design, and probabilistic systems. It reinforced that AI design has to start with human needs, and real transformation only happens when departments align around collaboration, governance, and trust.

Problem framing and principles → Started from real citizen pain points, then overlaid where AI could responsibly support staff and citizens. Set experience principles grounded in trust, inclusion, explainability, and human fallback. Research and ecosystem mapping → With limited user access, triangulated insights from public data, government reports, surveys, and social media discussions. Mapped the Irish public-service ecosystem (Justice, Housing, Education, Social Protection, HSE) to surface duplication, gaps, and handoffs. This highlighted where people need clarity and where fragmented data blocks proactive support. Proto-persona and use cases → Built the concept around a proto-persona, Martha, a newcomer to Ireland navigating visa, healthcare, education, and housing. Used her journey to define priority use cases and show how AI could reduce cognitive and admin load, while keeping fairness, privacy, and human oversight front and centre. Mapping and prototyping → Visualised the AI experience logic through journey maps and system diagrams, defining three progressive layers: Pre-Arrival Guidance (multilingual, trusted info), Personalised Support (tailored guidance across departments), and Proactive Assistance (predictive prompts with human review).

Impact

The proof of concept translated responsible AI principles into something tangible: end-to-end system journeys and orchestration logic for an agent-driven interface, plus a service vision and prototype for human-centred, AI-enabled public services. It also surfaced practical opportunities for cross-department data integration and governance, which are key enablers for trust at scale. The concept won the Dublin Regional Final of the Here for Good AI Hackathon 2025 and progressed to the national final, sparking new conversations with government clients and reinforcing Deloitte’s position in trustworthy AI design. Personally, it stretched my thinking on AI UX, conversational design, and probabilistic systems. It reinforced that AI design has to start with human needs, and real transformation only happens when departments align around collaboration, governance, and trust.

Previous — AI to accelerate R&D

Previous — AI to accelerate R&D

Previous — AI to accelerate R&D

Previous — AI to accelerate R&D

Previous — AI to accelerate R&D

Previous — AI to accelerate R&D

Next — Communicating AI impact

Next — Communicating AI impact

Next — Communicating AI impact

Next — Communicating AI impact

Next — Communicating AI impact

Next — Communicating AI impact

Enter password to view work