DataFYI

Datafyi is an agentic-based search platform built on the Walrus datastore. The concept will be expanded to allow agents to query the datastore and make payments via x-402. In this process, we have three agents, regardless of whether the query is made by an agent or a human:

1.Parser Agent: This agent uses our exposed API endpoint to convert a natural language query into tags. 2.Selector Agent: This agent selects which data matches the query based on specific criteria, returning the top "k" search results. 3.Payment Process: The user can then select the data they want to buy, and the curator of that data gets paid.

Additionally, we have defined an SDK that enables agents to connect to our data source and facilitate these operations.

Instead of a traditional search engine, our system enables agents to search based on tags and descriptions from our datastore. Users can access the data after payment via a BlobID from Walrus. Initially designed for AI agents to access and pay using x-402, this feature will evolve further with an SDK already in place.

We’ve built the platform using Next.js, Rainbow Kit, and Wagmi for wallet connections, Walrus for storage, PYUSD for payments, Langchain for agents, and Python with FastAPI to define agent endpoints. Notably, we've replaced the search engine with a step-by-step approach to facilitate web interaction for AI agents.