Deloitte Nexus Enterprise AI Agent Platform
Centralizing enterprise data through a multi agent platform
Nexus is an AI agent platform that simplifies enterprise data access and automates routine workflows for professionals in Government and Public Services.
I led the 0-to-1 design from concept to production, aligning product, engineering, and our primary target users. Through close co-design with this group, I shaped the first experience to support one of the most nuanced workflows in the domain.
Role
Design lead
Team
Product managers
Engineering teams (Frontend, Backend, and AI/ML)
Data scientists
Duration
April 2025 ~ Current
Impacts
42% reduction in contract setup time (saving an estimated 14,000 hours annually)
4.2 / 5 user satisfaction score post-launch
-
Accessing distributed enterprise data through simple conversation
Contract professionals struggled to find critical information across 50+ fragmented systems. I designed the conversational experience for Nexus so users could access distributed data through simple chat, reducing context switching and bypassing outdated interfaces of multiple systems.
-
Moving beyond queries to agentic workflows
The goal was not just to answer questions but to improve productivity by helping users complete their work. I designed the first end-to-end workflow for a high impact task, enabling agents to automate manual steps and guide multiple users through complex processes. This significantly reduced time spent on repetitive activities.
-
Building accuracy and trust through human in the loop transparency
Introducing agents required confidence in their outcomes. I created human validation points that make agent reasoning verifiable, including document-level citations, transparent thinking steps, and contextual checkpoints. These mechanisms allow users to confirm results, maintain control, and trust the agentic experience.
This project is available by request. Please enter the password you received or contact jongyeonchae@gmail.com to request access!