VANA - About us
The first network for user-owned data
We believe in an open internet where users own their data and the AI models they contribute to.
AI models should be created more like open source software: iteratively by a community. To make this possible, researchers need access to the world's best datasets that are held captive across walled gardens. Users can break down these walled gardens by exporting their own data.
We are building towards a user-owned foundation model, trained by 100M users who contribute their data and compute.
Vana's Platform
Vana aims to build a platform that lets users "pool" their data-chats, speech recordings, and photos-into datasets for generative AI model training. They also seek to create personalized experiences, such as daily motivational voicemails based on wellness goals or art-generating apps that understand style preferences, by fine-tuning public models on user data.
What's the motivation?
From Big Tech firms to startups, AI developers are licensing e-books, images, videos, audio, and more from data brokers to train more capable (and more legally defensible) AI-powered products. Shutterstock, for example, has agreements with Meta, Google, Amazon, and Apple to supply millions of images for model training, while OpenAI has deals with several news organizations to use their news archives.
In many cases, the individual creators and owners of that data haven't seen any of the profits. Vana wants to change that.
Founding of Vana
Anna Kazlauskas and Art Abal co-founded Vana in 2021 after meeting in a class at the MIT Media Lab focused on building tech for emerging markets. Before Vana, Kazlauskas studied computer science and economics at MIT and launched a fintech automation startup, Iambiq, through Y Combinator. Abal, a corporate lawyer by training, was an associate at The Cadmus Group before leading impact sourcing at data annotation company Appen.
Vana's API
The Vana API connects users' cross-platform personal data to personalize applications. This access allows apps to use a user's personalized AI model or underlying data, simplifying onboarding and reducing compute cost concerns. Vana believes users should be able to bring their data from platforms like Instagram, Facebook, and Google to create personalized experiences from the first interaction with a consumer AI application.
Creating an Account
Creating an account with Vana is simple. After confirming your email, you can attach data to a digital avatar (e.g., selfies, descriptions of yourself, and voice recordings) and explore apps built using Vana's platform and datasets. These apps range from ChatGPT-style chatbots and interactive storybooks to a Hinge profile generator.
Data Privacy Concerns
Given the current climate of increased data privacy awareness and ransomware attacks, why would someone volunteer their personal info to an anonymous startup, particularly a venture-backed one? Vana has raised $20 million from Paradigm, Polychain Capital, and others. Can any profit-driven company be trusted not to abuse or mishandle monetizable data?
Kazlauskas emphasized that Vana aims for users to "reclaim control over their data." Users can self-host their data rather than store it on Vana's servers and control how their data is shared with apps and developers. Vana makes money by charging users a monthly subscription starting at $3.99 and levying a "data transaction" fee on developers, disincentivizing the exploitation of users data.
"We want to create models owned and governed by users who contribute their data," Kazlauskas said, "allowing users to bring their data and models with them to any application."
Reddit
Reddit Data DAO
Vana is not selling users' data to companies for AI model training but wants to allow users to do this themselves if they choose, starting with their Reddit posts. This month, Vana launched the Reddit Data DAO (Digital Autonomous Organization), pooling multiple users Reddit data (including karma and post history) and allowing them to decide together how the combined data is used. Users can join with a Reddit account, request their data from Reddit, and upload it to the DAO, gaining voting rights on decisions like licensing the data to AI companies for shared profit.
This initiative responds to Reddit's recent moves to commercialize data on its platform. Reddit, which initially didn't gate access to posts for AI training, reversed this policy late last year before its IPO, earning over $203 million in licensing fees from companies like Google.
Reddit's Reaction
Reddit is not working with Vana officially and is displeased with the DAO. It banned Vana's subreddit and accused Vana of exploiting its data export system, which complies with regulations like GDPR and the California Consumer Privacy Act. A Reddit spokesperson stressed that their data arrangements include guardrails to prevent misuse, emphasizing that Reddit does not share non-public, personal data with commercial enterprises.
Vana's DAO Future
Kazlauskas envisions the DAO growing to impact the amount Reddit can charge for its data. However, with only 141,000 members out of Reddit's 73 million users, the DAO has a long way to go. Distribution of payments from data buyers is also a challenge. Currently, the DAO awards cryptocurrency tokens corresponding to users' Reddit karma, but karma may not be the best measure of data quality. Kazlauskas suggests members could share cross-platform and demographic data to increase the DAO's value and incentivize sign-ups, but this requires more trust in Vana's data handling.
Website: https://www.vana.com/
