AI-powered Content Generation & Item Condition Reporting for Premium Moves

Customer Overview

A UK-headquartered private equity firm acquired companies that provide premium services for moving, storing, and caring for high-value items of elite clients in the US and Europe. They aimed to enhance the accuracy and efficiency of move processes.

Project Overview

The client wanted to improve the accuracy of critical processes like item profiling and condition reporting conducted while moving the possessions of their high-net-worth clients. They wanted to help clients get visibility into the status of their items, both on the move and in storage.

Challenges

Building a digital solution to ensure detailed item profiling and accurate condition reporting while reducing time and effort. Providing customers complete visibility into their items with natural language search capabilities and multilingual support.

• The moving team must carry out detailed item profiling & condition reporting before moving items. Every project involves different types of items. However, the resources cannot write thorough descriptions of every item & accurately report their condition.
• Even with trained professionals & best practices, human error is inevitable. This can be costly, especially when critical defects in priceless possessions like family heirlooms are missed during reporting.
• As these professionals handle over 500 items per move, accurately documenting every detail of every item is time-consuming and tedious.
• Clients do not receive status updates on their items during move-ins and move-outs because the entire process operates offline.
• The client aimed to elevate service standards by offering item profiling and condition reports in their elite customers' regional languages, not just English.
• They wanted to enable their customers to search for their items using natural language on the company's Client Portal and obtain the status of those items.

Solution

We developed an AI-enabled iPad app for automated item profiling and condition reporting. Our team built multi-language reporting capabilities and enhanced item search to include natural language querying for customers in the Client Portal.

• We built an iPad app for the moving team to carry out the item profiling and condition reporting processes. The personnel takes high-resolution photos of each item and reviews Gen AI-generated titles, descriptions, tags, and condition reports before finalizing them.
• The moving team marks circles on items’ images to highlight areas needing attention during item profiling & condition reporting. They review AI-generated content & edit it. We used Claude 3.5 LLM Model to generate content & detect defects with humans in loop.
• We utilized Amazon Translate to make the reports available in the customers' preferred languages.
• Leveraging the AI-generated detailed and accurate titles, descriptions, and tags, we extended the existing item search functionality on the Client Portal. We used the RAG concept of Gen AI, Titan Multimodal Embeddings, and Claude 3.5 LLM to allow customers to search for their items in natural language.

Benefits

• The time taken by the moving team to create critical reports per item decreased from over 15 minutes to under 3 minutes.
• Our client digitized and automated move processes while achieving accuracy, transparency, and efficiency.
• Customers get more detailed and accurate item profiling and condition reports in multiple languages of their choice.
• Customers stay informed about the real-time status of their items on the move and in our client’s warehouses.

Technology

• Python
• LlamaIndex
• ChromaDB
• Claude 3.5 Sonnet Models
• Amazon Titan Multimodal Embeddings models
• Amazon Translate
• Amazon Bedrock

Industry

• Logistics
• Warehouse management