MoveSpace Overview
Decentralized datahouse for LLMs and blockchain use cases
MoveSpace is a Web 3 vector-based AI data engineering warehouse. It is an on-chain labeling protocol and a data marketplace, powered by a vector database. It can serve as the backend to power new Web 3 applications and use cases including advanced search engine, AIGC in gaming and social, AI-powered NFT, and LLM training.
In terms of content generation, MoveSpace can provide applications and users with ready-to-use text-to-text, text-to-image, image-to-text, and a continuously expanding multi-modal pipeline of content generation capabilities, based on MoveSpace's vector database.
Use case highlights
Data search engine (powered by vector data)
Developers and projects can upload data to our vector database via API and deploy our search engine SDK to embed a search feature in their app.
Examples include:
Quest campaign platforms can provide a search engine feature on their frontend to allow users to look up certain campaigns with human natural language prompts.
Data and NFT marketplaces can provide a search feature that allows users to look up specific content type with natural language. Such a prompt can be: give me the names of the NFT collections that have the largest new holder increase over the past 15 days; give me the datasets that can train image-to-text models that are uploaded over the past 2 days.
AIGC in gaming and social
Games that have storytelling elements or NPCs can upload their game script to MoveSpace's vector database via API and use MoveSpace's AI-powered dashboard to generate new scripts to power their in-game storytelling and NPCs. Players would be able to interact with NPCs in real-time and host dynamic conversations.
For fully on-chain games, MoveSpace's vector database can serve as the backend for generating in-game content.
AIGC in coding
Developers can easily upload smart contract coding language examples to MoveSpace's vector database via API. Based on the supplied code examples, users can use MoveSpace's AI-powered dashboard to generate new code to solve programming problems.
AI-powered NFT/characters
NFTs can be knowledgeable and intelligent with human input. Data and human prompts can be inserted into an NFT as metadata to empower NFT to act as autonomous agent in an environment.
Metadata can be further added to with human input to make the NFTs evolvable.
Labeled data for LLM training
Developers can launch data labeling campaigns on MoveSpace's data labeling dapp to call on questors to label datasets based on their instructions. Labels are created on BNNBChain/opBNB and datasets are stored on Greenfield with verifiable on-chain ownership. Labeled datasets can only be accessed and used by the original owner unless it is sold on BNBChain to someone else. Datasets can then be downloaded for LLM training.
Last updated