MinIO introduces AIStor Memory for agent long-term memory and secrets

MinIO said it released AIStor Memory for enterprise AI agents to keep long-term memory, workspace state, and secrets in a single integrated system. The company said the approach was designed to support repeatable agent work across interactions while keeping enterprise knowledge under the organization’s control.

According to MinIO, AIStor Memory preserves what AI agents learn across interactions so that knowledge stays discoverable and reusable to other authorized agents. The system is described as delivering relevant context to AI models to improve response quality while reducing latency and token costs, with knowledge secured, governed, and governed under the organization’s control.

MinIO said AIStor Memory functioned as a native data type available alongside objects and tables, rather than a set of separate components. It said the memory layer addressed patterns in the AI agent stack where teams otherwise assembled persistent memory by combining object storage, vector stores, metadata databases, secrets managers, and synchronization pipelines. The release described interfaces over HTTPS or a POSIX folder mount.

MinIO said the capabilities included memory scaling with storage rather than a fixed context window and a design that prevented truncation, summarization, or eviction. It also said the product used erasure coding, bitrot protection, encryption, compression, and tolerance to drive and data center failures. MinIO said memory stayed on infrastructure the customer owns under keys held by the customer, and the release listed use cases such as software engineering agents across large codebases, deep research and analysis spanning hours or days, and human-in-the-loop workflows that pause and resume over extended periods.

“Knowledge generated by AI agents becomes organizational memory, and organizational memory belongs on enterprise-controlled infrastructure,” said AB Periasamy, co-founder and CEO of MinIO. “AIStor Memory brings long-term memory, persistent workspaces, and secrets together on a single enterprise-controlled foundation. A single agent’s memory becomes the shared substrate for the entire organization.” “Secure execution and durable memory are both essential for production AI agents,” said Ivan Burazin, Co-Founder & CEO at Daytona. “Daytona gives agents an isolated runtime in which to work, while AIStor Memory preserves memory, workspaces, and secrets beyond the life of any individual sandbox. That separation allows compute to remain disposable while organizational knowledge stays durable and governed.” “As AI agents move from experimentation into production, memory is becoming an important part of the enterprise AI stack rather than an application feature. Enterprises need a durable, governed way to preserve context across sessions without creating another fragmented layer of infrastructure,” said Stephanie Walter, Practice Leader, AI Stack, HyperFRAME Research. “Integrating memory directly into the data platform can simplify deployment while giving organizations greater control over security, governance, and long-running agent workflows.” “Agentic AI cannot operate reliably at enterprise scale without durable, governed memory,” said Asher Lohman, CDO and SVP of Data & Analytics at Trace3.

Provided by Globe Newswire on behalf of MinIO. Click to read original content. The original article was written by Decision Insights Editorial.