Hey Product Hunt community! 👋 I built a portable SSD for AI Agents. 🤯 If you use AI daily, you know the "context amnesia" problem. Your knowledge is scattered across Gemini, ChatGPT, Claude, and Cursor. Every time you switch platforms, you lose continuity and have to start over from scratch. Cortex Protocol is the bridge. I built a universal memory layer on MCP that acts as a shared brain across every tool you use. The "Biological Funnel" Architecture: I didn't just build a database; I mimicked the human brain with three distinct layers: 🔴 Hot Layer: Fresh, immediate conversational context (last 48h). 🟠 Warm Layer: Summarized memories that evolve and "decay" (fade away) if they aren't used, keeping your brain efficient. 🔵 Cold Layer: Permanent, long-term knowledge storage. The Philosophical Flip: Most memory systems are built to help developers improve their apps. Cortex is built for the user. You decide what gets stored. You decide what gets shared. You decide what stays hidden. Where I am & Where I'm going: In this version (V1), I'm using a background "Dreamer" pipeline (powered by Gemini 3.1 Pro) to merge duplicates and resolve conflicts. It’s effective, but I'm already working on Cortex V2, which will support years of memory with near-zero degradation and bring costs down to under $1/month per user. My vision is simple: Cortex is designed to become the default memory layer for all AI agents... including robots in the future. 🤖 I'm excited to hear your feedback! What should your AI remember next? Happy building, David Fernandes

Cortex Memory MCP

Universal Long-Term Memory for AI Agents via MCP

Universal Long-Term Memory for AI Agents via MCP