EscrowIQ (Hackathon Project)
EscrowIQ is an intelligent assistant embedded within the escrow file experience, designed to automate activity tracking, surface insights, and reduce manual effort for escrow officers. It transforms fragmented logs and updates into actionable, real-time guidance.
The Problem
Escrow officers had to manually review fragmented activity logs to identify important changes, prioritize attention, and determine next steps.
Discovery & Research
Because this was a rapid hackathon project, the research focused on understanding the existing escrow activity experience, identifying repetitive information-seeking tasks, and determining where AI could provide the most value without disrupting the escrow officer’s primary workflow.
Research Goal
Understand how escrow officers review file activity and identify opportunities for AI to help them quickly answer three important questions: What changed? • What needs my attention? • What should I do next?
Research Approach
I focused the research on the existing escrow file and activity-log experience by:
• Reviewing how file activities and updates are presented.
• Mapping common information-seeking tasks within the activity timeline.
• Identifying repetitive actions and areas of cognitive load.
• Evaluating which activities could be summarized or surfaced by AI..
• Collaborating with engineering to understand data and API feasibility.
Key Findings
1. Activity logs provide history, but not insight.
2. Important changes can be difficult to identify.
3. Users need answers, not just activity records like who was added or removed, what requires immediate attention, and what actions should happen next.
4. Repetitive questions create an opportunity for AI
5. The assistant should remain contextual to the current escrow file and help users understand information while leaving important decisions and actions under the escrow officer's control.
Design Opportunity
Transform the activity log from a passive system of record into an intelligent source of guidance.
Design Implications
The research led to several design principles for EscrowIQ:
Contextual
AI should understand the current escrow file and its activity history.
Embedded
The assistant should live within the existing workflow rather than requiring users to open another application.
Action-oriented
Responses should highlight key changes, important information, and possible next steps.
Easy to initiate
Smart prompts such as Activity Today, Tasks Assigned to Me, Generate Work Summary, and Next Steps reduce the need to formulate questions from scratch.
Human-controlled
AI should assist with interpretation and recommendations while keeping the escrow officer responsible for critical decisions.
Design Opportunity
An AI assistant embedded in the escrow file, offering contextual summaries, activity insights, task updates, and suggested next steps.
Impacts
Here are the measurable improvements observed