Illinois Legal Aid, AI-Enabled Platform

Challenges

Improving access to legal help online presents unique barriers for organisations that serve individuals unfamiliar with legal systems. Most of the time, users don’t realise what kind of legal problem they’re facing. As a result, ILAO’s keyword-based search was not effective for them.
ILAO previously offered live chat services, but the costs and ROI weren’t sustainable. Duplicate and misclassified content reduced reliability, making it harder for users to find actionable information. Accuracy added another layer of complexity. The goal in building the AI-assistants was to ensure that responses reflect current Illinois law and draw from verified material.  Off-the-shelf models were combined with a RAG approach, using a custom chunking strategy to ensure responses drew only from verified Illinois legal resources. Additional safeguards and guardrails were implemented to maintain accuracy and compliance.

Approach

We began by observing how people searched for legal help and where they struggled. User research revealed the limitations of keyword searches and the effort required to find relevant information without prior legal knowledge.
Rather than applying a general-purpose model, we chose a retrieval-based approach anchored in ILAO’s verified legal content. 
This allowed responses to remain grounded, traceable, and easier to interpret. Design and engineering moved in parallel. 
User behaviour shaped interaction patterns, and search refinement was designed to reduce friction. 
Time and compliance constraints shaped every layer of the system. The result is a shift in how users interact with legal information: more guided, consistent, and aligned with how people naturally ask for help.

Solutions

We began by observing how people searched for legal help and where they struggled. 
User research revealed the limitations of keyword searches and the effort required to find relevant information without prior legal knowledge. Rather than applying a general-purpose model, we chose a retrieval-based approach anchored in ILAO’s verified legal content. This allowed responses to remain grounded, traceable, and easier to interpret. 
Design and engineering moved in parallel. User behaviour shaped interaction patterns, and search refinement was designed to reduce friction. 
Time and compliance constraints shaped every layer of the system. The result is a shift in how users interact with legal information, more guided, consistent, and aligned with how people naturally ask for help.

Semantic search

A custom-built semantic search engine replaced ILAO’s keyword-based system. It interprets user intent, ranks responses by legal relevance, and handles ambiguity through prompt-based refinement, drawing from ILAO’s vetted legal content. This improved the accuracy of search results and made legal information easier to find for time-pressed users.

Conversational legal assistant powered by RAG

An AI-driven assistant simplifies access to legal help by providing structured, plain-language responses through a conversational interface. It eliminates the need to navigate multiple pages and supports users with low digital literacy by offering clarity in complex legal scenarios. 
Behind the scenes, a domain-specific retrieval-augmented generation (RAG) architecture ensures that every response is grounded in ILAO’s trusted legal corpus. This setup ensures compliance with Illinois law while delivering accurate, scalable guidance across a wide range of legal topics.

Backend and infrastructure modernisation

An AI-driven assistant simplifies access to legal help by providing structured, plain-language responses through a conversational interface. It eliminates the need to navigate multiple pages and supports users with low digital literacy by offering clarity in complex legal scenarios. Behind the scenes, a domain-specific retrieval-augmented generation (RAG) architecture with a custom chunking strategy ensures that every response is grounded in ILAO’s trusted legal resources. Layered prompting techniques guide the model to pinpoint the exact information needed, while guardrails verify that answers are accurate, complete, and compliant with Illinois law. This setup ensures reliability and transparency while providing accurate, scalable guidance across a broad range of legal topics.

Outcome

ILAO’s rollout of conversational AI assistants and semantic search has helped it serve more users and better deliver on its mission as a legal aid organisation.
‍
1. Growth in users reached
‍Search page sessions increased from around 100 to nearly 5,000 - a sign that more people are finding and using the improved search experience.
‍
2. Better interaction with legal information
‍Users now get direct answers to their questions instead of clicking through multiple pages. This helps them find relevant legal content more quickly and easily.