overview
Introducing AI to a conservative, high-risk enterprise environment
In early 2023, I was tasked with the end-to-end UX and front-end delivery of an internal AI assistant for a national law firm. The goal was to provide lawyers and staff with a secure environment to summarize documents and draft correspondence without compromising confidential data.
I joined the project when the system existed primarily as backend logic. As the sole designer and front-end engineer for the interface, I was responsible for translating complex model capabilities into a functional, production-ready web application.
Role
Lead UX Engineer • Product Design • Front-end Architecture
Users
Lawyers and professional legal staff
Tech Stack
ASP.NET Core, jQuery, Vanilla JS, CSS3
problem
Bridging the trust gap in high-stakes environments
While the backend was technically capable, the initial interface lacked the polish and responsiveness required to gain user trust. In a legal environment, perceived reliability is as critical as technical accuracy; if the tool felt "buggy" or slow, adoption would remain low.
Primary Challenge
Modernizing the interaction patterns of a legacy tech stack to meet the real-time expectations of generative AI.
implementation
Standardizing complexity through technical restraint
Key Deliverables
• Asynchronous UI: Engineered a front-end state machine to handle token streaming and "thinking" indicators, mitigating perceived latency.
• Information Architecture: Developed a system for conversation persistence, including time-based grouping and favoriting for repeat legal workflows.
• Systematic UI: Built a modular set of chat components within the existing legacy framework to ensure consistency and rapid iteration.
• Constraint Management: Maintained strict adherence to enterprise security protocols regarding data handling and prompt visibility.
The focus was not on competing with consumer-facing features, but on creating a predictable, secure tool tailored specifically to the firm’s internal data and confidentiality requirements.
Context
• Legacy front-end environment
• No formal research phase
• High-security data constraints
• Rapid development cycles
outcomes
Validating demand through enterprise adoption
Following deployment, the tool became the 2nd most used application, becoming a central component of the organization’s digital workflow.
Performance Impact
The tool's success demonstrated the viability of AI in the legal space, leading to the establishment of a dedicated AI-focused internal team.
Institutional Growth
The project successfully transitioned from a technical proof-of-concept to a primary firm-wide investment.
walkthrough
Detailed technical review
Due to the sensitive nature of the data involved, I can provide a private walkthrough of the codebase, technical architecture, and UI components upon request.