.png){kind=small}
AI-OCR System
Project Type:
Timeline:
Status:
Overview:
5 Weeks
The EAGLYS AI-OCR project was created for a leading aircraft manufacturer struggling with massive volumes of technical manuals, schematics, and repair documents in various formats (paper, scans, PDFs).
AI
B2B
Done
Project Background
Aircraft Maintenance Documentation Digitization
Project Outcomes
-
Hundreds of documents digitized and indexed in the first week of rollout.
-
Client reported search tasks that previously took hours are now completed in seconds.
-
Quickly adopted by engineers, including senior, non-technical staff.
-
Improved workflow efficiency and laid groundwork for future expansion across maintenance teams.
The EAGLYS AI-OCR project was developed for a leading Japanese aircraft manufacturer to streamline the handling of vast technical manuals, repair logs, and diagrams. Engineers previously spent hours searching across fragmented PDFs and folders.
The new system digitized these assets into an AI-powered, searchable database — cutting retrieval times from hours to seconds and allowing the maintenance team to focus on what truly mattered: fixing aircraft.
The Problem
Engineers at this major Japanese aircraft manufacturer were drowning in fragmented documentation, physical paper records, manuals, repair logs, and schematics scattered across paper folders and PDFs.
Retrieving information was:
Slow
Time wasted manually searching through unindexed folders.
Error-prone
Inconsistent naming and manual tagging caused confusion.
Inefficient
Valuable knowledge locked in unstructured formats.
“Finding a single part number or repair reference could take anywhere from 15 minutes to a full day”, slowing maintenance and introducing costly delays.
The Goal
Digitize and organize thousands of maintenance documents into a centralized, searchable database using AI-powered OCR to improve efficiency, accuracy, and accessibility for the engineering team.
We proposed an AI-OCR system to turn the fragmented manual process into a single streamlined pipeline, from scanning and classification to AI-based analysis and search. What once required multiple tools and manual tagging can now be achieved with a few clicks.
My Role & Approach
Sole product designer collaborating with one AI engineer, one full-stack developer, and a project manager.
Vision
-
Defined the vision with the project manager, aligning design goals with client needs.
-
Validated feasibility of design concepts through close collaboration with AI and full-stack engineers
UX Improvements
-
Proposed model tier naming (“Standard,” “Easy,” “Heavy”) + tooltips to clarify tradeoffs.
-
Advanced settings to give users more control over analysis parameters and fine-tuning.
-
Tags for improved searchability and file management.
UI Redesign
-
Rebuilt the PoC interface into a clean, file-based dashboard with consistent hierarchy and searchable structure.
Delivery
-
Handed off wireframes, user flows, and high-fidelity prototypes for implementation.
Key Screens
Home screen
Before: Engineering-led PoC design with limited usability.
After: Redesigned as a familiar file-management dashboard with search, quick upload, and easy browsing of documents.
Upload & AI Analysis
Before: The name of the models had no context and meaning for the user.
After: Streamlined flow where users select a document, choose a pre-trained model (added a tooltip to add context and meaning) and let the system handle preprocessing automatically with the option of advanced settings.
File View & Term Finder
Before: Basic viewer with limited navigation.
After: Full-page view with thumbnails, searchable tags, term highlighting, and a side panel to navigate OCR results, making searches intuitive and efficient.
The Results
Efficient
Faster task turnaround and fewer errors when referencing technical manuals
Functional
Search and tagging improved discoverability dramatically
Intuitive
Familiar file-system interface, which reduced training time and adoption friction
Feedback source: Pilot testing phase, 2023
「新しいAI-OCRシステムのおかげで、エンジニアが重要な情報を見つける作業が大幅にスピードアップしました。システム内の文書を扱うのも簡単で、直感的に操作できます。」
“The new AI-OCR system helped speed up the process of finding important information for the engineers. It’s simple to use and intuitive to deal with the documents inside the system.”
My Reflection
-
Design clarity builds trust: Even complex AI processes need to be translated into intuitive, familiar interfaces for non-technical users.
-
Early collaboration is key: Working closely with engineers from the start prevents rework and ensures technical feasibility.
-
Structured layouts build trust: Clear organization and consistent UI patterns help users feel confident navigating large, complex datasets.
