Glossary

Address

A blockchain address (usually a hexadecimal identifier) representing a user’s account or a smart contract. In DeepBlock, addresses are treated as entities in the knowledge graph. They can hold assets, initiate transactions, and interact with other addresses.

Indexer

A component or service that continuously scans blockchain data and organizes it for querying. Indexers connect to blockchain nodes, process new blocks and transactions, and store the results in a structured database. DeepBlock’s indexers are the foundation of its data ingestion, allowing for efficient retrieval of on-chain data without hitting raw RPC endpoints for every query.

Knowledge Graph

A database that represents information in a graph structure of nodes (entities) and edges (relationships). DeepBlock’s knowledge graph models the blockchain world (addresses, transactions, tokens, etc.) as a network of connected data. This enables intuitive queries like “how is X connected to Y” or “what’s the path from A to B,” reflecting real-world relationships on-chain.

GraphQL

A query language for APIs and a runtime for fulfilling those queries with your data. Unlike REST, GraphQL allows clients to request exactly the fields they need and get all related data in a single request. DeepBlock’s API is GraphQL-based, which means you can craft queries to navigate the knowledge graph and retrieve structured JSON results containing precisely the on-chain data points you asked for.

RAG

Retrieval-Augmented Generation. An AI technique that combines a retrieval step with generative AI. “Retrieval” means fetching relevant reference information (from a database, document store, etc.), and “Generation” refers to an LLM generating an answer or content. By augmenting the generation with retrieved facts, RAG ensures the output is grounded in up-to-date and specific data. DeepBlock uses RAG so that AI agents can answer questions about blockchain data by first pulling the latest facts from the knowledge graph.

MCP

Model Context Protocol. A protocol or set of guidelines for formatting context given to an AI model. In DeepBlock, MCP dictates how on-chain data and related context should be organized when inserted into an LLM’s prompt. The purpose is to maintain clarity and structure, so the model can easily understand the data and its relationships. MCP is what makes DeepBlock’s AI integration effective. It’s why an AI agent can accurately interpret multiple blockchain facts without getting confused.

TVL

A common DeFi metric standing for Total Value Locked. It represents the total value of assets (usually in USD) that are locked or staked in a smart contract or DeFi protocol. For example, if a lending protocol has many users depositing funds, the sum of all those deposits (converted to USD) is its TVL. TVL is often used to gauge the popularity or health of a DeFi platform. DeepBlock’s knowledge graph tracks TVL across protocols and can provide historical and real-time TVL values as part of queries.

DeFi

Short for Decentralized Finance, an ecosystem of financial applications built on blockchain networks (primarily Ethereum and similar platforms). DeFi applications include decentralized exchanges, lending platforms, stablecoins, yield farming, and more, all of which operate without centralized intermediaries. DeepBlock focuses heavily on DeFi data (transactions, liquidity pools, governance, etc.), making it easier to analyze this fast-moving domain.

LLM

Large Language Model. A type of AI model that’s been trained on vast amounts of text to understand and generate human-like language. Examples include GPT-4 and other transformer-based models. LLMs can answer questions, summarize information, and perform reasoning tasks. In DeepBlock’s context, an LLM is used to interpret user questions and generate explanations or insights, with the help of retrieved blockchain data (via the RAG pipeline). Essentially, the LLM is the “brain” that provides narrative or answers, and DeepBlock is the “eye” that provides the facts for it to look at.