• 

  MindVault Dashboard — Privacy-focused AI wellness companion interface

MindVault — Private AI Wellness Companion

Inspiration

Personal wellness conversations and journaling can contain some of the most sensitive information a person creates. We wanted to explore whether an AI wellness companion could provide useful reflection and support without requiring users to give up control over their private data.

This led to MindVault, a privacy-focused AI wellness companion combining AI-powered conversations, mood tracking, private journaling concepts, and Midnight's privacy-preserving blockchain technology.

What it does

MindVault is designed as a private space where users can:

• Record and reflect on journal entries
• Track their mood over time
• Interact with an AI wellness companion powered by Gemini
• Use a privacy-focused interface for sensitive personal information
• Integrate wellness activity with Midnight's Compact smart-contract environment

The project focuses on demonstrating how AI and privacy-preserving blockchain technology can work together rather than treating personal wellness data as ordinary public application data.

How we built it

MindVault was built by extending Midnight's example-hello-world DApp and adapting it into a wellness-focused application.

Midnight integration

The project includes a Compact smart contract:

• contracts/mindvault.compact
• A recordJournal() circuit
• Private journal-related state handling
• Midnight local devnet integration
• Contract compilation and automated testing

The Compact contract provides the privacy-oriented blockchain component of the application, while the application layer handles the user experience and AI interaction.

AI layer

Gemini powers the AI wellness companion, allowing users to interact with an AI assistant for reflection and wellness-oriented conversations.

The AI component is intentionally separated from the blockchain layer: sensitive conversational content does not need to become public blockchain state simply because the application uses a blockchain.

Application layer

The project includes a web interface for interacting with the MindVault experience, together with the Midnight wallet/provider integration required to interact with the local network.

Challenges

One of the biggest challenges was combining several technologies that each have different development requirements.

We had to work with:

• Midnight Compact smart contracts
• Zero-knowledge circuit compilation
• Midnight's local development network
• Wallet and provider integration
• TypeScript application code
• AI API integration
• A privacy-focused user experience

Another challenge was understanding what information should actually be placed on-chain. A key design principle was to avoid treating the blockchain as a conventional database for sensitive wellness conversations.

What we learned

Building MindVault gave us practical experience with Midnight's privacy model, Compact smart contracts, local network tooling, wallet integration, and zero-knowledge-oriented application development.

We also learned that privacy is not simply a feature added at the end of an application. It needs to influence the architecture from the beginning: what data is collected, what is processed by the AI, what is persisted, and what is exposed to a blockchain.

What's next

MindVault is currently a prototype demonstrating the integration of AI wellness interactions with Midnight's privacy-preserving technology.

Future work could include:

• Stronger end-to-end private journal storage
• More advanced mood and wellness analytics
• Improved AI personalization
• User-controlled encrypted data storage
• More extensive Midnight testnet deployment
• Additional privacy-preserving circuits

MindVault explores a simple question:

Can an AI wellness companion be useful without requiring users to sacrifice privacy?

Built With

• compact
• docker
• gemini
• javascript
• midnight
• node.js
• typescript