engrammemory-ai

Engram Memory is persistent AI memory for agents that need to remember conversations, decisions, and context without storing data on someone else’s servers. It is presented for use with coding agents, customer support, healthcare, multi-agent systems, edge deployments, and IoT, and the site says it works with Claude, ChatGPT, Gemini, Python, and Node.js through OpenClaw Agents SDK.

Its core functions are described as API Intelligence, Overflow Storage, Proprietary Compression, and a Cross-Platform Bridge. The API turns text into something that can be searched by meaning rather than keywords and avoids saving duplicate information. Overflow Storage keeps older memories when local space runs out and returns them later, using automatic tiering between local hot storage and cloud warm storage with opt-in encrypted-at-rest handling. The compression feature is said to reduce memory to one-sixth of its size. The Cross-Platform Bridge is described as end-to-end encrypted sync between self-hosted instances so memories can be shared across devices without a central data store.

Engram says it processes data in transit and then sends it to a vector database, with the data staying on the user’s infrastructure. It describes itself as self-hosted, local-first, and capable of full offline operation. The site also states that the core is open source under the MIT license.