Patent Filed for Reference File Management for Artificial Intelligence Models

Strategic Insights -

According to a patent filed on February 13, 2024, and published online on October 14, 2025, a system has been developed to optimize reference file management for artificial intelligence (AI) models. The system's primary goal is to improve the performance and efficiency of AI models by efficiently storing and retrieving reference files. This innovation could significantly enhance the capabilities of AI models, particularly in applications that rely on vector databases, such as language models and computer vision systems.

In the current state of AI model development, reference files can consume a substantial amount of storage space, often necessitating the use of lower-performance memory to keep costs manageable. This can result in significant delays when users or applications request access to reference files related to a specific response. The patented system addresses this challenge by integrating a data storage device (DSD) with advanced features, including the ability to prioritize the storage of reference files, create cache zones for efficient retrieval, and associate reference files with their respective vector embeddings.

This system has the potential to revolutionize AI model efficiency, enabling them to process and respond to complex queries more quickly and accurately. The efficiency of reference file management in the patented system can also lead to cost savings, as it reduces the memory required to store reference files.

Key Takeaways:

  • A patent has been filed for a system that optimizes reference file management for artificial intelligence models.
  • The system aims to improve the efficiency and performance of AI models by efficiently storing and retrieving reference files.
  • The patented system utilizes a data storage device (DSD) that integrates advanced features, including prioritized storage of reference files, cache zone creation, and association of reference files with vector embeddings.
  • The system has the potential to revolutionize AI model efficiency, enabling faster and more accurate processing of complex queries.
  • Efficient reference file management can lead to cost savings by decreasing the memory required to store reference files.
  • The system's developments can significantly enhance the capabilities of AI models, particularly in applications relying on vector databases, such as language models and computer vision systems.
  • The patented system has the potential to improve AI model accuracy, especially in tasks that involve complex queries and require referencing multiple files.
  • The system can also enable more effective use of computing resources, thereby reducing the need for additional infrastructure or resources.
  • A total of 20 claims have been made to further explain the patented system, including five short method claims and 15 system claims for data storage devices.

Statistics:

  • The patent was filed on February 13, 2024, and published online on October 14, 2025.
  • The patented system includes features such as prioritized storage of reference files, cache zone creation, and association of reference files with vector embeddings.
  • The system aims to improve AI model efficiency, enabling faster and more accurate processing of complex queries.
  • Efficient reference file management can lead to cost savings by decreasing the memory required to store reference files.
  • The system has the potential to revolutionize AI model efficiency and accuracy, especially in applications that rely on vector databases.

Sources:

  • U.S. Patent Number 12443534
  • Patent URL (for desktop use only): https://ppubs.uspto.gov/pubwebapp/external.html?q=(12443534)&db=USPAT&type=ids