Legal Assistant

Fragmented tools, duplicated work

Legal staff switched between CMS, SharePoint, Microsoft Word, email, and printed documents just to complete one contract — with no automation or AI support at any step.

Incomplete requests caused back-and-forth

Requesters frequently submitted the wrong contract type or missing information, forcing the team to deny and restart — often losing multiple days per case.

No visibility into workload or status

The approver manually allocated work with no tracking system. Neither legal staff nor requesters had clear visibility into where a contract was in the process.

Research

Interviews & Observation

Conducted 1-on-1 sessions with legal staff, approvers, and requesters across the organization. Observed real drafting sessions from start to finish — watching how users moved between CMS, SharePoint, Word, and physical documents — to capture the full workflow and uncover friction that users themselves hadn't articulated.

Synthesis

Affinity mapping & Journey analysis

Clustered hundreds of raw observations and interview notes into themes using affinity mapping. Built user journeys and workflow models for all three user groups — legal staff, approver, and requester — to identify where the most critical pain points and drop-off moments occurred across the contract lifecycle.

Strategy

Feature ideation & Scoping

Facilitated individual and group brainstorming sessions that generated 9 AI feature concepts — ranging from auto-fill and document classification to PDPA data blur. Led the team through a prioritization exercise to evaluate each idea against user needs, technical feasibility, and business value before defining the product scope.

Design

Wireframes & Prototyping

Translated research insights into user flows, lo-fi wireframes, and high-fidelity prototypes in Figma for desktop and mobile devices. Designed separate interfaces for legal staff and requesters, covering key features such as AI-assisted contract drafting, inline spell-check with Accept/Dismiss controls, auto-fill from request forms, and a document checklist system.

AI contract drafting
Chat-style input lets staff describe what they need; AI generates a draft from the matching template.

Document classification
Scans uploaded attachments, classifies type, and checks against the required document checklist.

PDPA data blur
Automatically detects and blurs sensitive personal data in uploaded images before processing.

Auto-fill from request
Maps data from the requester's form directly into template fields — reducing copy-paste errors.

Spell & grammar checker
Inline error detection with Accept / Dismiss controls, categorized by error type.

AI report query
Natural-language queries let users pull summaries and reports from contract data on demand.

See how I approached product discovery and AI concept research →

Legal Assistant

Smarter Contract Drafting

The problem