Lumina

Lumina is the ultimate all-in-one mental health tool, combining therapy, journaling, and dreamsculpting to address real-world issues. The app allows users to log significant life events and dream experiences, utilizing AI to generate images and scenes from those dreams. Users can interact with customizable AI therapists available 24/7, choosing their gender, voice, and personality traits. These therapists can identify connections between childhood experiences, life events, and dream themes, linking them to therapy discussions. Lumina analyzes dreams, revealing themes and potential meanings behind various elements. Additionally, the app offers audio packs that guide dreams in specific directions while users sleep, including dream storylines, subliminal messaging, and AI-generated sleep audio. Users can create dreamsculpting stories with specific themes and names, and can simply fall asleep to these narratives. The app's features extend to guided dream sculpting with planned audio that plays at certain points during sleep, sleep tracking, meditation, and horoscope integration. In the "About Yourself" section, users can share their life stories and incorporate horoscope themes, receiving weekly or daily horoscope readings linked to their personal experiences. Lumina addresses significant issues: it provides accessible therapy without the high costs of live sessions, requiring no insurance, and allows users to engage in therapy on their own schedules, eliminating the need for pre-scheduled appointments. It also supports individuals dealing with unresolved childhood trauma, nightmares, and night terrors, with plans to include a meditation feature. Lumina is designed to solve the pressing issue of data sensitivity in the medical world by ensuring users have control over their information. Sensitive data is encrypted and securely stored, addressing the widespread concerns of data breaches in mental health services. By offering a solution that prioritizes user ownership of data, Lumina stands out as a safer alternative to competitors like BetterHelp and Aura. With the potential for infinite use cases and monetization strategies, Lumina represents a transformative step forward in mental health support. After deploying this product, the world can expect enhanced privacy in mental health care, greater accessibility to therapy, and improved overall well-being for users. Lumina also features a unique data pool option, allowing users to voluntarily share their anonymized data to discover others facing similar challenges. This feature enables the AI to connect users with like-minded individuals, fostering a sense of community and support. By engaging with others who have shared experiences, users can gain valuable insights and encouragement on their mental health journeys. To enhance user engagement, Lumina includes daily prompts that encourage users to reflect and journal regularly. By responding to these prompts, users can earn rewards that unlock new features and functionalities within the app. This gamified approach not only motivates consistent journaling but also promotes healthy habits. After answering their daily prompts, users have the option to meditate, further supporting their mental well-being and enhancing their overall experience with the app.

Demo Video here: https://youtu.be/cqBXG5GA5Hk?si=L5s--c8qCgYGzzX4

1. AI Therapy with Omi AI We started by integrating Omi AI for the core functionality of AI therapy. Omi AI provides a conversational interface that adapts to users’ mental health needs through NLP and sentiment analysis. The plugin we developed for Omi AI acts as a middleware to handle user sessions, logging therapy interactions, and allowing dynamic responses based on the user's input.

Tech Stack for Omi AI Integration: Node.js with Express: We built the middleware using Node.js to handle incoming requests from the front-end and route them to Omi AI’s API. Express allows for scalability and flexibility in managing these requests. Next.js: For our front-end, we use Next.js, which communicates with the Omi AI plugin through API routes for SSR and dynamic loading of the AI’s responses. This enables fast interaction without reloading pages and allows us to easily manage different states (such as ongoing therapy sessions).

1. Data Encryption with Lit Protocol After users log their therapy sessions, journals, or dreams, the raw data is immediately encrypted using Lit Protocol. Lit Protocol is a decentralized access control protocol that ensures sensitive user data can only be accessed with the proper decryption keys. Here's how it works in detail:

Encryption Process: AES Encryption: Once the data is logged, we encrypt it using AES-256 in the browser before it even reaches our servers. This guarantees that the data is secure during transit. Lit Protocol Access Control: We use Lit Protocol's SDK to wrap the encrypted data in a decentralized access control layer. Users control the access keys and can decide who can decrypt their data.

Ethereum Wallet Integration: User authentication for encryption/decryption is handled via an Ethereum wallet connection. The private key of the user’s wallet plays a key role in controlling access, ensuring that no one without the user’s permission can view the encrypted data.

Tech Stack for Encryption: Lit Protocol SDK: Integrated within the Next.js app, allowing for seamless encryption and key management directly from the front-end. Ethereum Wallet (MetaMask): Users connect their wallets, which control the encryption/decryption process via Lit Protocol.

1. Data Storage via Walrus/Sign Once the data is encrypted, it’s passed to storage protocols like Walrus or Sign. These decentralized storage solutions offer both scalability and security. We implemented these protocols to ensure that user data remains off centralized servers and is distributed in a way that only authorized entities can access it.

Storage Process:

Walrus/Sign IPFS Integration: Data is stored using IPFS (InterPlanetary File System) through Walrus/Sign. The encrypted data blobs are uploaded to IPFS, with a unique hash generated for each file.

Decentralized Access Management: Only users with the correct encryption keys can retrieve and decrypt the data from IPFS. This decentralized storage approach minimizes the risk of data breaches, as no single entity holds the complete unencrypted dataset. Tech Stack for Storage:

IPFS (via Walrus/Sign SDKs): IPFS is used to store the actual encrypted data. These SDKs ensure that files are correctly pinned and retrievable. Ethereum Smart Contracts: Some of the access control logic is managed via smart contracts, ensuring that data retrieval and permissions are auditable on the blockchain.

1. Messaging System with XMTP For user-to-user communication, especially for those who wish to connect anonymously over shared experiences, we integrated the XMTP (Extensible Message Transport Protocol). This decentralized protocol enables end-to-end encrypted messaging across wallets.

Messaging Process:

Anonymity: Users communicate using wallet addresses, ensuring that their real identity remains anonymous. XMTP enables messages to be linked to a user’s Ethereum wallet without revealing their actual identity. Opt-In Communication: Users can opt in to communicate with others who have logged similar dreams or experiences. We built a match-making algorithm that identifies overlapping themes and triggers potential connections between users. Message Encryption: XMTP ensures that all messages are end-to-end encrypted. Even our servers never have access to the plaintext messages. Tech Stack for Messaging:

XMTP SDK: Integrated directly into the Next.js app to manage message encryption, decryption, and wallet-based communication. Next.js API Routes: Used to handle the initialization of messaging sessions, establishing connections between users based on shared themes.

1. Hacky/Notable Aspects Hacky Storage Encryption Workflow: To make encryption ultra-fast, we implemented client-side encryption before data hits the server. By using the Crypto API directly in the browser, we can ensure that even if the connection is compromised, the data is still inaccessible. Decentralized Messaging System: We had to hack around the XMTP SDK’s message storage and retrieval systems, making sure users could send encrypted messages that remain invisible even to us, the developers. This was done using a combination of local message queuing and XMTP routing.

Data Deletion Protocol: We created a custom protocol where users can request to fully delete their encrypted data from all nodes storing it, leveraging smart contracts to track the deletion process. We ensure that once deleted, the data’s encryption keys are also revoked, making the data permanently inaccessible.

Conclusion This project is a powerful combination of decentralized storage, end-to-end encryption, and AI-powered therapy. By using Omi AI, Lit Protocol, Walrus/Sign, and XMTP, we’ve built a secure, privacy-first platform where users can safely engage in therapy, log personal experiences, and connect with others anonymously without compromising their data.