Concept
Legal AI
Exploring AI-assisted legal search through clearer IA, search logic, and trust-building interaction patterns.
Role
Product Designer
Domain
Enterprise AI / legal search
Focus
IA, search logic, trust-building interaction patterns
01 The Setup
Context & Problem
Legal professionals spend a reported 40 to 60% of billable time on document research. Existing search tools require precise keyword matching, which sits badly with the semantic nature of legal language.
Making AI-assisted semantic search feel trustworthy and auditable to a profession defined by precision and liability. Every AI-surfaced result had to be traceable, explainable and verifiable. This stayed a concept: nothing here was built or put in front of a law firm.
02 Design Decisions
What We Tried & Why We Pivoted
Relevance provenance
Every AI-surfaced result shows why it was surfaced: which clauses matched, what semantic concepts connected it, and how similar documents rank nearby.
Side-by-side comparison
Fluid two-panel comparison of documents with synchronized scrolling and AI-highlighted parallel passages.
Saved research threads
Research sessions persist as structured threads: queries, results, annotations and decisions logged chronologically for audit and collaboration.
03 Outcomes
Results
Concept
design exploration, not a shipped product
04 Reflection
What I'd Do Differently
Explainability is not a feature you add on top. If a reader cannot understand why the AI surfaced a result, they will not use it regardless of accuracy.