Insight is a decentralized emotion data labeling network.
Insight is a decentralized emotion data labeling network designed to power multimodal Emotion AI. Users upload selfies, label their emotions, and earn non-transferable on-chain tokens as proof-of-work. These human-labeled emotion datasets are used to train our in-house multimodal Emotion AI model, which is already integrated into two Tarta Labs products — SpotZero (AI anime game) and omo (AI emotional companion).
Insight leverages Web3 incentives to scale global emotion data collection, ensuring higher diversity, authenticity, and volume compared to traditional annotation pipelines. In the future, Insight will provide an Emotion AI SDK for builders, enabling any application to integrate emotion understanding capabilities.
During this hackathon, we implemented the on-chain reward module for $mEMO on Mantle, enabling users to receive non-transferable tokens after completing emotion labeling tasks. We also integrated a verification pipeline that links user contributions to their on-chain records, deployed a prototype of the Insight web dashboard, and connected our Emotion AI model to the labeling workflow. This version demonstrates the full flow: user → emotion capture → annotation → AI validation → on-chain reward → dataset generation.
Tarta Labs is funded by top-tier investors such as Bitkraft, Spartan, and Hashkey, and supported by Mantle and Arbitrum grants. Insight is an early-stage product developed within Tarta Labs and has not raised its own independent funding yet. Our next fundraising round is now in preparation.