Update My Brain is Brian Swichkow's shorthand for the interaction he wants between 📝Claude and 📝MythOS: one plain conversational instruction, carried out by a system that already holds the user's context and knows which workflow applies.
The phrase names a target experience, not a single feature, and its aim is minimal friction: the user should not have to engineer prompts over and over, organize context by hand or build each workflow from scratch. Deciding what to gather and how to proceed moves from the person to the system.
The friction it answers is the manual upkeep of current AI workflows. Brian points to people maintaining "50 CLAUDE.md files" to carry project context across sessions, each a 📝CLAUDE.md instruction file written and kept current by hand. He describes that state as fragmented, tiring and dependent on repeated 📝prompt engineering, a cost that grows for people working across many projects and domains.
In the intended model, MythOS holds structured context and Claude reasons over it, connected through the 📝MythOS MCP. Asked to prepare a work session on a topic, the system gathers the relevant memos, people, concepts and templates into a project-ready context, so the user no longer searches and pastes material in. Within 📝Memory → Mind → Mouth → Myth, Claude acts as mind and mouth over MythOS memory. The companion idea of a 📝process memo applies the same principle to sharing workflows between people.
The sentence I keep returning to is this one: "I want you to just be like, update my brain, and it knows what to do."