index.as

UseCase1-ComposableEcosystem
UseCase2-Retail-Brand+Taste
UseCase3 - Journalism-Different Perspective
GIF
composable.prototype

Inspiration

Our inspiration for Index.as came from the need to decentralize information discovery in web3. While data had become decentralized, discovery remained centralized, relying on traditional search engines and platforms. We wanted to empower creators and individuals to build their own contextual discovery engines that interconnect with each other, enabling a more decentralized and inclusive discovery ecosystem.

What it does

Index.as allows creators to create contextual discovery engines using their own information. It enables data ownership, collaboration, and monetization to form a discovery ecosystem as a network. It cultivates greater participation in decentralized discovery and creates a diverse and inclusive ecosystem. We use NFTs as creator roles for broader contexts with large groups. Index.as is useful for brands, creators, communities, DAOs, curators, researchers, and enthusiasts.

How we built it

Here is the detailed information about the architecture: https://github.com/indexas/indexas In the context of the Augment Hackathon:

We use ComposeDB to leverage the decentralized data network and enable interoperability among indexes.
Thanks to Ceramic, we store data on IPFS in a decentralized manner, ensuring availability and resilience.
To manage access control within the indexes, we utilize NFTs with the LIT Protocol.
In order to enable composability among indexes, we used LlamaIndex and Langchain. We have created a ComposableGraph that allows for querying multiple indexes.

Challenges we ran into

We couldn't optimize for performance due to the time limit of the hackathon, which limits the user experience.
While we have enabled composability, in order to provide a meaningful experience, we understand that we should provide instructions to indexes using LLM Prompt templates.

Accomplishments that we're proud of

We have built the world's first truly decentralized search and discovery network.
We have demonstrated that composability can be enabled, providing new directions in information discovery, thanks to interoperability.

What we learned

What we learned when hacking: We started by enabling composable queries over multiple discovery engines but we discovered there could be additional composability layers such as multi-user conversations within discovery engines. What we learned from the community:

We can embed cryptographically verifiable AI Models thanks to Modulus Labs
We could embed long-term and short-term memory within our apps. (Thanks to Richard from Algovera.ai)
We've come across an exciting new way to generate revenue for our project: "Pay per query."

What's next for index.as

We will integrate HuggingFace to enable algorithmic choice for discovery engines.
Our queries feature, which allows for easy composition, will be moved into production.
We’ll enable creator monetization for discovery engine creators.