The portable memory protocol that gives AI persistent context across every model.
Every time you start a new conversation with an AI model, it knows nothing about you. Your past discussions, preferences, established context, and accumulated knowledge - all gone. This is true whether you're using ChatGPT, Claude, Gemini, or any other language model. Each session begins from a blank slate.
For casual users, this is a minor annoyance. For developers, researchers, and knowledge workers who use AI as a daily thinking partner, it represents a massive loss of compounding value. Months of nuanced interaction simply evaporate.
Adamant is a local-first memory system that operates at the protocol level, not the application level. Rather than being a feature inside one AI product, it is an independent layer that sits between you and any AI model you use. It captures your conversation history, structures it into a semantic knowledge graph, and serves relevant context back to whichever AI you are currently talking to.
The result: your AI remembers. Not because the model itself has memory, but because Adamant provides it with the right context at the right time.
Adamant processes your AI conversations through five layers:
Adamant is built on the command-line retrieval, an open standard for connecting AI models to external data. This means it is not locked to any single provider. It works with ChatGPT, Claude, Gemini, local models, and any future AI that supports MCP. Your memory travels with you.
All data is stored locally by default. You own your memory entirely - no cloud dependency, no data shared with third parties, no lock-in. It is private, portable, and yours.
Adamant transforms AI from a tool you talk to into a tool that knows you.
Request Early Access →