Archipelago is an AI-powered platform that helps enterprise clients manage and share complex property insurance data. As a Product Designer, I contributed to several core features including geotagging tools and global filtering systems. This case study focuses on a new file management experience built to organize thousands of insurance documents across client accounts, which I delivered quickly and without access to traditional search functionality.
I led the UX and UI design of the file management MVP, collaborating closely with product managers, engineers, and client-facing teams. The project required strategic design decisions within a narrow technical scope and tight timeline. My goal was to create a system that would deliver clear value to users despite significant infrastructure limitations.
The primary users were property specialists and account managers who regularly worked with large volumes of insurance documentation. Many of them previously used Dropbox, Google Drive, or internal workarounds to store and share files. These generic solutions often slowed down critical workflows and didn’t reflect how insurance files are actually used.
Our team was tasked with designing and shipping a file management MVP that could:
Organize thousands of documents across multiple client accounts
Surface relevant files quickly using metadata and AI-assisted tagging
Increase platform value by replacing the need for third-party storage tools
However, search functionality was unavailable due to backend limitations. File names were often inconsistent or unclear, which ruled out basic string-matching or naming-based filters. The design needed to compensate for all of this and still deliver a fast, usable experience.
I began by partnering with account managers to understand how they located and shared documents under pressure. Through interviews, live working sessions, and recorded calls with clients, I uncovered their tagging strategies, naming workarounds, and the mental shortcuts they used to retrieve specific file types.
Based on these insights, I designed a system focused on high-clarity filtering, consistent metadata, and intuitive categorization. Each file could be tagged by document type, date relevance, and usage context. The system supported flexible combinations of filters so users could isolate key files, even in large or inconsistently named data sets.
I also introduced subtle UI patterns that nudged users toward better naming hygiene. Even though automation wasn’t available, these small interventions helped drive more consistent file structures over time.
We launched the MVP internally with account managers, who frequently needed to retrieve documents like SOVs under time-sensitive conditions. Prior to the redesign, finding a single file could take anywhere from 1 to 12 minutes. After the launch, the average retrieval time dropped to 30 seconds, with a maximum of 2 minutes.
This shift significantly improved daily workflows, and validated that UX even without a search engine can drive both speed and perceived product quality.