A LLaMA2-7b chatbot with memory running on CPU, and optimized using smooth quantization, 4-bit quantization or Intel® Extension For PyTorch with bfloat16.

Embedded long-term memory database for AI agents & LLM apps. Six-layer cognitive architecture, knowledge graph, hybrid BM25 + vector search, sleep-time memory consolidation. One .meh file, zero infrastructure, pure Go.

AI chatbot that remembers users across sessions. OpenAI GPT-4o + Next.js + persistent memory. Clone and deploy in 5 minutes.

Lightweight 3-tier memory and context management for AI agents. Token-budget-aware, SQLite-backed, zero-config. Works with Claude, OpenAI, or any LLM agent system.

Drop-in long-term memory for LLM apps — application-layer Memory Caching (Behrouz et al. 2026, arXiv:2602.24281). 14.6x fewer tokens at higher accuracy. 3-line API.