AI Powered Proxy Archiving

Digital media and data assets such as photos, documents, audio, and video files have for decades been stored on secondary devices for long-term retention. Many of these formats and devices are obsolete or no longer supported by the vendor. For example, Sony announced that it will “gradually cease production of optical disc storage media products, including Blu-ray discs.” Not only is the hardware becoming obsolete, but so is the software used to playback the media. These secondary devices used for long-term storage have software and operating systems (e.g., Windows 95) that are now decades old and no longer supported by the original manufacturers.

With hardware that can no longer playback media, data, applications, an OS that cannot run the software to display the assets, and long-term storage solutions for medical records, financial documents, and digital evidence becoming useless; the pressing question is, “How to migrate from these outdated secondary devices to an open platform and more future-proof solution?”

At Dynamic Workflow Solutions we tackled this problem for multiple verticals by developing an AI-powered Proxy Archiving Solution powered by our Asset Migration Service (AMS). The software handles hundreds of discs at a time via robotic devices with on-premise processing. AMS scans through folders and files to parse metadata, which can come from file names or sidecar files (e.g., XML, JSON, CSV), or both. It also converts proprietary video files to a standard MP4 format at the binary level and standardizes metadata to prepare for the next step: AI processing.

Once standardized, the media assets and metadata are ready for AI processing. The AI enriches the metadata, enhancing search criteria and significantly reducing discovery time by cross-linking information. It also stores corresponding data chunks in a vector database for later retrieval, utilizing Retrieval-augmented generation (RAG).

RAG is a software architecture that combines the capabilities of large language models (LLMs), known for their broad knowledge, with specific business information from documents, SQL databases, and internal applications. RAG enhances LLM response accuracy and relevance by storing up-to-date information in a vector database, which is retrieved using semantic searches and added to the prompt context. This process enables the LLM to provide the best possible, up-to-date responses. This architecture is renowned for its ability to offer detailed, relevant answers by integrating the extensive knowledge of LLMs with proprietary business data.

The enriched metadata, original media asset, and working MP4 file are sent to Hybrid for the final archiving step. Hybrid generates a proxy preview of all media assets, storing it locally

?

while parsing the enriched metadata into the database. Simultaneously, the original media is sent to cold storage, Deep Archive, ensuring cost-effective storage with maintained security and redundancy. By creating the proxy and keeping the metadata in the database, everything is searchable including video, audio, images, documents full context search, cell phone extractions and more. Users can review the Proxy Media Assets, share them across the platform, create new collections or cases, and decide whether to rehydrate a single file or the entire case.? By using Proxy Archiving the high cost of Egress and Ingress is dramatically reduced because the user can function with the media and data prior to large “rehydration operations” to view their assets.