Architecture at a High Level

Project Zero's architecture integrates data ingestion, semantic enrichment, knowledge graph construction, and AI Agent execution into a single coherent pipeline. This document provides a comprehensive overview of the system architecture and its components.

System Description

Project Zero is designed to process blockchain data through a sophisticated pipeline that enriches, normalizes, and makes data available for various use cases. A typical data flow begins with raw blocks ingested from multiple blockchains and proceeds through stateless block functions, which normalize and annotate transaction data. This enriched information is then routed into Super Indexers for efficient query handling, while also populating a knowledge graph that represents addresses, tokens, contracts, and their relationships.

Core Components

Stream Processing Architecture

The stream processing architecture consists of several key components:

1. Block Stream: The entry point for blockchain data, receiving gRPC raw data from various networks.
2. Block Function: A sandboxed environment where user code executes to filter and transform the raw data.
3. Sink: The output component that delivers processed data to different destination types (Webhook, S3, Postgres, Kafka, etc.).

Supporting components include:

• State Manager: Observes and manages the stream state
• Telemetry: Collects logs and metrics for monitoring

System Overview

The Project Zero system operates with these primary components:

1. Block Streams: Dedicated streams (shown in pink) for each blockchain network
2. Stream Processing: Core processing units (shown in green) handling data transformation
3. Cluster Manager: Orchestrates stream operations
4. Control Plane Platform: User interface for stream management via API
5. Destination: Various supported output destinations (shown in yellow)

Use Cases

AI Agent Integration

Project Zero can be used to build AI Agents with the following components:

1. Project Zero Core: Provides the base data ingestion
2. Database (DB): Stores processed blockchain data
3. Providers: Interface with blockchain services
4. Actions: Execute blockchain transactions
5. Evaluators: Assess opportunities and risks

The AI Agent can query the knowledge graph and Zero APIs to gather real-time context for its decisions. When the agent identifies an opportunity or triggers a predefined logic condition, it initiates an on-chain transaction or a series of transactions.

Zero API Implementation

The Zero API implementation showcases another use case with the following components:

1. Project Zero: Initial data ingestion point
2. DataLake: Central storage for processed data
3. Query Engine: Facilitates data access and analysis
4. Zero API: Provides external API access

Technical Details

Stream Processing

Each stream in Project Zero follows a specific processing flow:

1. Data Ingestion:

• Receives raw blockchain data via gRPC
• Validates data integrity
• Queues blocks for processing

2. Block Function Execution:

• Runs in isolated sandbox environment
• Applies user-defined transformations
• Implements filtering logic
• Enriches data with additional context

3. Sink Delivery:

• Supports multiple destination types
• Ensures exactly-once delivery
• Handles backpressure
• Manages connection pools

State Management

The State Manager provides:

• Stream state tracking
• Checkpoint management
• Recovery mechanisms
• Configuration versioning

Monitoring and Telemetry

The system collects:

• Performance metrics
• Processing latency
• Error rates
• Resource utilization
• Custom metrics from Block Functions