MythOS is a knowledge-infrastructure platform built by 📝Brian Swichkow — "infrastructure for thought" that unifies personal memory, contextual intelligence, and expressive output in one system. It runs on a knowledge graph of interlinked memos, ships an MCP server and an AI augmentation framework, and layers communities and email ingestion on top. Founded in 2017 in Los Angeles.
The platform was AI-native years before the term existed. Brian started it in 2017 and put the first AI agent, BrianBot, over its memory that same year — an agent acting on a structured personal knowledge base, not a chatbot bolted onto a notes app. The MCP server is the load-bearing piece: it turns the graph into a knowledge base that external tooling can query directly, so memory becomes something agents read from and act on rather than something a human scrolls through.
I had Brian on the pod to dig into his 📝cold-email and digital-empathy methods, and MythOS is the system underneath — where his context lives so his agents can act on it. For a GTM engineer the interesting bet is the MCP server: a knowledge base your outbound tooling can query, not just a notes app.