Block Whisperer

Screenshots

Problem Statement

Using Blockchain explorers is sooo 2005.Its time to upgrade the experience of finding out about onchain activity!BlockWhisperer using the power of LLMs to allow users to search on onchain activity using natural language.But this isn't your grandmas AI search. Instead of just showing text, it creates Generative UI elements based on the search results to create an enhanced User Experience.Here is how it works:Step 1:User asks a question. Example "What was Lucs's Last Mint?"Step 2:The Large Language Model (GPT-4o) takes this question and decides what functions/tools are needed to answer this question.Step 3:The functions are then run. In this example:Firstthe LLM will lookup an ENS related to Luc.Step 4:The LLM will present a Generative UI element to the user with Lucs's ENS name on it to get confirmation from the user.Step 5:An API request will be sent using the Wallet Address of Kartik to theZerion API. This will include theoperation_typebeing mint.Step 6:Then the LLM will Genearte a UI element to the user with the transaction data of Luc's last mint.What Can You AskLookup the Wallet Address for an ENSFind the social accounts of an ENSFind the NFTs Owned by a ENSFind the Crypto Owned by a ENSFind the transactions by a ENSFind a specific transaction by an ENSFind details on a certain token

Solution

Technologies UsedWeb3 RelatedZerion API- Used for the getting data about wallets including crypto owned, nfts owned and transactions.ENS Lookup- Used to lookup ENS names to wallet addresses that are required for API calls. Also used to get social accounts of an ENS.Coinmarket Cap API- Used to get the stats of the a give cryptocurrency.Non-Web3GPT4o- The Large Language model powering the natural language experience.Vercel AI SDK- Used to create Generative UI elements. This gives an enhanced user experience instead of dealing with text.Tool Invocation- Through the Vercel AI SDK, the LLM can make tool calls and run functions to get answers for users.