Second Me is an open-source project from Mindverse that lets you train a personal “AI self” — a model that captures your knowledge, context and way of thinking, then represents you to others. It has more than 15,600 stars on GitHub and was most recently updated in 2025.
Second Me is built on Mindverse’s published research into AI-native memory, a memory architecture designed to store and retrieve personal context the way the model itself represents it. You train your Second Me locally on your own data, so your information stays under your control rather than inside a company’s cloud. Once trained, it can answer in your voice, carry your context into conversations, and connect to a wider network of other people’s AI selves, scaling your presence without scaling your hours.
The technical bet behind the project is that a personal AI should be trained, not just prompted. Second Me’s training process builds what Mindverse calls a “second brain” from your notes, documents and other personal data, using its AI-native memory research — including a second-generation paper — to organize that material for retrieval and reasoning in your style. Owners can then chat with their AI self, put it to work as a stand-in that shares their context with collaborators, and plug it into the project’s emerging network where different people’s Second Mes can interact on their owners’ behalf.
Second Me is free and open source under the Apache-2.0 license. Training and running it locally means your only costs are your own hardware or compute.
Creators, consultants and professionals who want a digital twin of their expertise, and personal-AI enthusiasts who prefer owning their model and data.