Atulya started as memory infrastructure for agents and teams. But a durable brain needs more than storage and retrieval. It also needs safe memory evolution. That is why Atulya now supports Git-like memory repos with branches, commits, rollback, and fork-to-bank workflows.
Storing memory easy. Pile of facts. Searchable. Done.
But pile of facts is not intelligence. Intelligence is knowing what matters in the pile. What is hot right now. What is fading. What just spiked abnormally. What contradicts something believed six months ago.
That is BRAIN. Layer inside Atulya that thinks about what got retained — continuously, in background, without waiting to be asked.
Agent forget. That core problem.
Not dramatic failure. Quiet one. Agent help Anurag refactor module today. Tomorrow — same agent, same repo — ask about that module. Agent say "I don't have context for that." Like nothing happened.
Worse: agent make change. Confident. No memory of constraint set three sessions ago. Break thing. No warning. Just breakage.
Agent remember fact. "Anurag reviewed PR #88." Good.
Agent remember another fact. "anurag@company.com approved deploy." Good.
Agent asked: "What did Anurag do this week?" Agent say: "I found one PR review. Nothing else."
Wrong. Two facts. Same person. Agent did not know. Looked like two separate entities. Completely unrelated.
This is entity fragmentation. Not edge case. Default state of any system that stores facts without resolving who they are about.
Build a ChatGPT-style chatbot with persistent memory using the OpenAI SDK and Atulya. Three API calls — retain(), recall(), reflect() — and your app remembers users across restarts, no vector database or RAG pipeline required.
AI agents forget everything between sessions. Atulya gives them persistent, structured memory via MCP. One Docker command to run the full stack locally. Connect any MCP-compatible client. Three core operations: retain (store), recall (search), reflect (reason) — plus mental models that auto-update as memories grow.
atulya-litellm adds persistent memory to any LLM provider via LiteLLM — OpenAI, Anthropic, Groq, Azure, Bedrock, Vertex AI, and 100+ more. Three lines of setup, and every LLM call automatically gets context from past conversations.
CrewAI agents lose all memory when a crew finishes. atulya-crewai plugs into CrewAI's ExternalMemory to persist knowledge across runs -- three lines of setup, and your agents automatically store task outputs and recall relevant context.
@eight-atulya/atulya-chat adds persistent memory with one wrapper function