GraphMySelf
A personal AI memory layer that syncs your identity across every AI tool and model you use.
The Problem
Every AI tool you open starts from zero. It does not know your projects, your experiences, your writing style, or how you think. You end up re-explaining yourself hundreds of times a year across ChatGPT, Claude, Cursor, and every other tool you use. It is repetitive, it breaks your flow, and it makes AI feel generic even when it should feel personal.
The Solution
GraphMySelf builds a semantic knowledge graph of who you are from the sources that already define you. Connect your resume, GitHub, LinkedIn, and Notion once. From that point on, every AI tool you use knows you.
Key Features
Personal Knowledge Graph Your identity is parsed, chunked, and stored as vector embeddings in a semantic graph. The system understands relationships between your skills, projects, experiences, and context, not just keywords.
Visual Graph Dashboard An interactive node-based UI built with React Flow lets you see exactly how you are represented. You can edit any node and the system re-embeds it instantly to keep your context accurate.
Works Across Every Model GraphMySelf is model-agnostic. Whether you use ChatGPT, Claude, Gemini, or anything else, your memory layer works the same way across all of them.
MCP Server Integration GraphMySelf exposes an MCP server so AI tools like Cursor and Claude Desktop can query your identity natively in real time, with no copy-paste required.
How It Works
You connect your data sources once. GraphMySelf parses everything using Apache Tika, runs it through an async job pipeline via BullMQ, generates vector embeddings via Google Vertex AI, and stores them in PostgreSQL with pgvector. At query time it retrieves the most semantically relevant chunks of your identity and passes them to whichever model you are using.
Impact
GraphMySelf removes the single most annoying friction point in daily AI usage. You stop being a stranger to your own tools. The more you add to it, the more useful every AI interaction becomes.
Currently in active development. Version 1 ships with the resume pipeline and MCP server support.