Descriptions:
One of the core frustrations with AI assistants is that every new session starts with a blank slate — no memory of past decisions, preferences, or context. In this video, Nate B Jones extends his OpenBrain concept, a personal memory system built on Supabase and connected to AI models via an MCP server, by showing how to give that memory system a practical, actionable interface for both humans and agents.
The central architectural idea is using a Supabase database table as a shared surface: AI agents access it through MCP to query, write, and update structured data, while users access the same data through a lightweight visual layer — a simple web page or mobile bookmark. Jones calls this the “human door.” There’s no additional middleware or separate platform; the table remains the single source of truth with two different doorways into it.
The video walks through six use cases built on this foundation, including household knowledge management (storing paint colors, appliance records, and Wi-Fi passwords as they come up in conversation), professional relationship tracking to surface neglected contacts, and proactive scheduling intelligence. Each use case follows the same pattern: Claude or another frontier model captures structured facts through ordinary conversation via MCP, while the visual layer adds search, categories, and at-a-glance organization over the same data. This is a practical follow-up for anyone who has already built an OpenBrain instance and wants to move beyond text-only chatbot interactions.
📺 Source: AI News & Strategy Daily | Nate B Jones · Published March 13, 2026
🏷️ Format: Hands On Build
