Doppelgangers AI

Doppelgangers.ai lets users create digital AI twin by using their data from messengers, emails & social media with ability to tokenize and share twin's personality traits on the Aurora blockchain

Comment 1

Inspiration

Black mirror - White Christmas, User Owned AI and digital immortality in a philosophical and utilitarian sense all have inspired the creation of Doppelgangers AI. Dopplegangers.ai is an innovative solution that reimagines an individual’s online presence, creating unique AI powered digital twins – Doppelgangers. Doppelganger is built using data produced and owned by user, controlled by user, and unbiased by the world. This is not a Large Language Model. This is not owned by big tech. This is true User Owned AI model.

What it does

Users create a digital AI twin version of themselves by uploading their Instagram account data (Facebook, TikTok and twitter coming soon), or linking email and messaging apps such as Telegram (WhatsApp, WeChat, Discord, or Slack coming soon) this creates an AI model that reflects user's unique personality and communication style.

Based on the digital twin users can develop custom multipurpose AI chatbots and integrate them within other applications using APIs (we only have telegram right now).

The datasets used to create digital twin can then be tokenized and turned into nfts. These can be shared on Aurora and/or NEAR blockchain allowing monetization and the feature we call Merge, where 2 users can merge their Digital twins into a new entity which will have characteristics of both users.

How we built it

The app has 2 main components - AI and blockchain. We are using python for user data processing, Chatbot kit for inference, backstory generation for each user based of that data (this is basically a digital twin component) and chat creation.
We are using smart contract interface, functions, events, and storage layout for the on-chain component of the app. The contract interacts with off-chain logic, which involves signing user-related metadata with a hidden private key generated by the app. The app can connect a web3 user wallet and use the same on chain contract’s interaction as for web2 users.

Challenges we ran into

Data encryption and user privacy

Accomplishments that we're proud of

Merge of 2 different user datasets to create a new model, design of our app.

What we learned

How to do on device inference.

What's next for Doppelgangers AI

NFT marketplace for digital twins created by users, celebrities and KOLs, on device inference for training of digital twins on user's smartphone for maximum privacy and security, create a RAG model for digital twins optimization, launch on NEAR main net, Go to market and TGE

Built With

Share this project:

Updates