Trustless Pilot

Trustless Pilot is an SDK that allows marketplaces or small businesses to integrate our composable review system.

The problem that we aim to solve is that right now product reviews garner very little credibility as they are mostly unverified and individuals often have an incentive to give dishonest reviews.

We aim to solve this problem in two ways: reputation and personalisation. Due to the composable nature of web3, a reputation-based reviewing system creates a greater incentive for individuals to review products and generates more accurate reputation scores since these are based on a larger data set. On top of this generic reputation by which review trustability is weighted, we also incorporate a personalised experience. We do this by accessing a user's social graph and weight the reviews of people closer in the social graph by a higher factor. Finally, we also create verifiable and secure proofs of purchase and humanity, both of which are positively related to the trustworthiness of a review.

Our MVP that we have built during this hackathon uses three different contracts: our own which stores reviews and review ratings (which represent how useful individuals find a particular review, the smart contracts of Lens Protocol to access individuals' social graphs and the Worldcoin contract to determine whether a particular Lens user is a human. We interact with all these contracts through subgraphs, some of which we built ourselves and others which are available online. These data points are collated in our backend, which calculates review ratings based on reputation and social graph proximity and returns different orderings of reviews for specific products depending on who looks at them. Finally, our SDK consumes this API and creates an interface allowing marketplaces and small businesses to easily integrate our product into their existing infrastructure.

Our tech stack is quite varied, using javascript, solidity and graphql as well as nextjs, expressjs, hardhat, mongodb and the graph, among others. We definitely surprised ourselves with what we built since the architecture is quite extensive and computationally expensive, which required us to continually optimise our code.

Our most hacky escapade was definitely trying out and ultimately discarding 5 different types of databases, one of those was a globalised caching system.