Vana, an EVM-compatible blockchain built around user-owned data, on Wednesday launched a beta app upgrade that lets people store personal data on their own device and carry their AI "memory" from one chatbot to the next.
The upgrade pairs a personal data server with a "Portable Memory" MCP endpoint and a Data Portability API, and is anchored by Vana's acquisition of the team behind Memory Protocol, according to a company blog post. Founder Jack Spallone has joined Vana to lead the effort.
Users connect sources once, from Spotify history to Oura sleep data to calendars, and the data stays local rather than on a platform's servers, Vana said. Any MCP-compatible tool, including ChatGPT and Claude, can read it with permission that's revocable anytime.
Vana joins a growing number of blockchain projects trying to pry AI memory out of corporate walled gardens. ZetaChain's Anuma touts a "private memory layer" that keeps context encrypted across models, PIN AI is building on-device assistants with on-chain data ownership, and NEAR Protocol calls its version "sovereign AI."
Vana framed the release as a counter to a wave of 2026 memory features from OpenAI, Google, Anthropic and Microsoft that keep user context on company servers. The beta is live at app.vana.org, with data-portability actions paid in the VANA token.
(HOW AI WAS USED IN THE PRODUCTION OF THIS PIECE: The first draft of this story was written in Claude's Cowork mode using a custom skill designed to handle press-release rewrites. I then used our Distro Publisher MCP connector to file the draft to DistroVerse, and I used the native story-editing interface to edit the piece. I personally checked all the facts prior to publication. I used Claude Cowork to create a story illustration using a custom-built skill.)