Everpure Adds AI Data Management and Inference Tools

Everpure announced in a press release new data management, inference and storage capabilities for enterprise AI workloads. The company says its PureKVA feature can reduce time to first token by up to 20 times by loading context directly into GPU memory.

Everpure Data Intelligence can discover, classify and add context to information across the Everpure Platform, public clouds, SaaS applications and third party storage. Native support for Model Context Protocol lets AI agents and security tools query live data catalogs using natural language, while file intelligence shows access permissions and stale data without reading file contents.

The updates also include DeepReduce, which continuously scans storage blocks for similarities to expand usable FlashBlade capacity. A reference architecture based on open weight models is intended to reduce external API token use and make AI costs more predictable. Everpure plans to make the capabilities available in October.