Making AI 'Remember Like Humans': Open-Sourcing "Cerememory," a Living Memory Database Based on Brain Mechanisms
Co-R-E Inc. has open-sourced "Cerememory," a brain-mechanism-based "living memory" database designed to give AI agents human-like memory. This aims to solve the challenge of long-term memory for AI.
📋 Article Processing Timeline
- 📰 Published: May 7, 2026 at 20:55
- 🔍 Collected: May 7, 2026 at 12:01
- 🤖 AI Analyzed: May 7, 2026 at 12:47 (45 min after Collected)
Co-R-E Inc. (Headquarters: Shinjuku-ku, Tokyo; Representative Director: Masato Okuwaki; hereinafter, Co-R-E), an AI connective company, has released "Cerememory," a memory layer that implements brain structures to provide AI agents with a "living memory" database. It is open-source and available to everyone.
Official Website: https://co-r-e.github.io/cerememory-docs/ja
GitHub: https://github.com/co-r-e/cerememory
■ Why refer to the human memory system?
Currently, developers worldwide are grappling with the challenge of giving AI agents long-term memory. Various excellent implementations have already been released, including memory functions from different companies, RAG using vector databases, and various context engineering techniques.
However, when surveying these efforts, one question arose: "When considering the ideal state of AI memory, what would be the optimal outcome?"
The answer to this question was actually very close: a living memory system, like that of us humans.
Human memory is not merely a mechanism for storing and retrieving information. Important experiences remain vivid, while trivial events fade. Similar memories blend, and recalling one triggers a chain of related memories. Everyone has experienced childhood memories resurfacing due to a nostalgic scent.
During sleep, the day's events are organized and integrated into long-term knowledge. It is an extremely sophisticated information processing system, refined over hundreds of thousands of years of evolution.
The question Cerememory addresses is not "how to give AI memory," but "to what extent can the excellent memory mechanisms acquired by humanity over a long time be incorporated into AI's memory layer?" Cerememory's approach is to implement phenomena observed in neuroscience research as an execution system, rather than merely as data structures.
■ Three Design Principles of Cerememory
Treat memory not as something "statically stored" but as something "dynamically alive."
Human memory is not statically stored like a hard drive. Important things are strengthened, unused ones fade, and they change slightly each time they are recalled. Cerememory inherits this characteristic, implementing dynamic processes such as decay, interference, reactivation, the organization of memories during dreams in sleep, and integration, as an execution system rather than merely as data structures. Memory is not fixed the moment it is written; it continues to move dynamically over time.
Memory should have "reasons" as well as "content."
Many people have probably struggled to understand the intentions behind an AI agent's actions. When a person remembers something, that memory is accompanied by meta-context such as "why it was considered important" or "how that decision was made." Cerememory incorporates a meta-memory plane into every AI agent's memory record, structuring and recording "why it exists." Intentions, rationale, evidence, alternatives, decisions. Memory can be traced not only by the content of the AI agent's actions but also by their reasons.
Make the memory layer independent of AI.
Human memory belongs to the individual and is not subservient to a specific conversational partner. Cerememory takes the same stance. Through its unique protocol "CMP*," it can access the same memory layer from any LLM, including Claude, GPT, and Gemini. Data is stored locally and can be exported at any time. Even if you switch AIs or cancel a service, your memory remains with you. The memory layer should not be owned by a specific AI or vendor.
*Note: CMP and MCP are different protocols. Although their names are similar and often confused, CMP (Cerememory Protocol) is a proprietary protocol defined by Cerememory for reading and writing memory, while MCP (Model Context Protocol) is an open specification proposed by Anthropic for connecting LLMs with external tools. The two are not competing; in Cerememory, "MCP carries CMP." This means that MCP clients like Claude Code or Cursor deliver CMP messages internally to the Cerememory engine.
■ "5-Store Architecture" that directly incorporates brain structure into its design
The human brain processes different types of memories in separate regions.
Official Website: https://co-r-e.github.io/cerememory-docs/ja
GitHub: https://github.com/co-r-e/cerememory
■ Why refer to the human memory system?
Currently, developers worldwide are grappling with the challenge of giving AI agents long-term memory. Various excellent implementations have already been released, including memory functions from different companies, RAG using vector databases, and various context engineering techniques.
However, when surveying these efforts, one question arose: "When considering the ideal state of AI memory, what would be the optimal outcome?"
The answer to this question was actually very close: a living memory system, like that of us humans.
Human memory is not merely a mechanism for storing and retrieving information. Important experiences remain vivid, while trivial events fade. Similar memories blend, and recalling one triggers a chain of related memories. Everyone has experienced childhood memories resurfacing due to a nostalgic scent.
During sleep, the day's events are organized and integrated into long-term knowledge. It is an extremely sophisticated information processing system, refined over hundreds of thousands of years of evolution.
The question Cerememory addresses is not "how to give AI memory," but "to what extent can the excellent memory mechanisms acquired by humanity over a long time be incorporated into AI's memory layer?" Cerememory's approach is to implement phenomena observed in neuroscience research as an execution system, rather than merely as data structures.
■ Three Design Principles of Cerememory
Treat memory not as something "statically stored" but as something "dynamically alive."
Human memory is not statically stored like a hard drive. Important things are strengthened, unused ones fade, and they change slightly each time they are recalled. Cerememory inherits this characteristic, implementing dynamic processes such as decay, interference, reactivation, the organization of memories during dreams in sleep, and integration, as an execution system rather than merely as data structures. Memory is not fixed the moment it is written; it continues to move dynamically over time.
Memory should have "reasons" as well as "content."
Many people have probably struggled to understand the intentions behind an AI agent's actions. When a person remembers something, that memory is accompanied by meta-context such as "why it was considered important" or "how that decision was made." Cerememory incorporates a meta-memory plane into every AI agent's memory record, structuring and recording "why it exists." Intentions, rationale, evidence, alternatives, decisions. Memory can be traced not only by the content of the AI agent's actions but also by their reasons.
Make the memory layer independent of AI.
Human memory belongs to the individual and is not subservient to a specific conversational partner. Cerememory takes the same stance. Through its unique protocol "CMP*," it can access the same memory layer from any LLM, including Claude, GPT, and Gemini. Data is stored locally and can be exported at any time. Even if you switch AIs or cancel a service, your memory remains with you. The memory layer should not be owned by a specific AI or vendor.
*Note: CMP and MCP are different protocols. Although their names are similar and often confused, CMP (Cerememory Protocol) is a proprietary protocol defined by Cerememory for reading and writing memory, while MCP (Model Context Protocol) is an open specification proposed by Anthropic for connecting LLMs with external tools. The two are not competing; in Cerememory, "MCP carries CMP." This means that MCP clients like Claude Code or Cursor deliver CMP messages internally to the Cerememory engine.
■ "5-Store Architecture" that directly incorporates brain structure into its design
The human brain processes different types of memories in separate regions.