The management of asset inventories in the field has become a challenge for many companies, especially for electricity providers who rely on manufacturers’ labels to accurately track their resources. In order to address this issue, an innovative artificial intelligence-based application has been developed that promises to significantly optimize this process.
The solution uses generative artificial intelligence technology and large language models (LLMs) to automatically extract relevant data from asset labels. This strategy aims not only to speed up inspections, but also to reduce human errors that often occur in manual data collection.
The system is based on Amazon Web Services (AWS) such as AWS Lambda and Amazon DynamoDB, allowing field technicians to take pictures of asset labels. Through advanced image analysis techniques, the application can automatically identify elements such as serial numbers and supplier brands. This capability is enhanced by a database that includes examples of similar labels, significantly improving the accuracy of extracted information.
The workflow is simple: the operator captures the image of the asset with their mobile device. Once the photo is uploaded to a secure storage, the system generates vector representations of the image and looks for matches in its database. Then, an LLM model is used to extract data. The accuracy of this information is verified by the technician, who confirms if the extracted data is accurate. If the system has high confidence in the accuracy, the information is automatically integrated into the inventory. On the other hand, data that does not meet the confidence threshold is sent for manual review.
This approach not only improves efficiency in asset inventory management but also allows operators to focus on other critical tasks, easing their workload. Additionally, by enabling more precise and efficient tracking of assets, organizations can ensure high-quality inventory records, minimizing the risk of errors that can arise during manual data entry.
With this solution, a path is opened towards the digitization of processes in industrial sectors that require rigorous tracking of their resources, standing out as a model for technological adoption in supply and asset management.
via: MiMub in Spanish
