OrchestratoR

An R-based orchestration system that coordinates AI agents to produce structured analytical reports through a modular set-based architecture.

System Overview

The OrchestratoR system transforms analytical requests into structured HTML reports through a carefully orchestrated process of set creation, verification, and assembly. At its core, the system uses a modular approach where each analysis component is created as a verified, self-contained set before being combined into the final report.

Core Components

Analysis Sets

The fundamental unit of analysis in the system is a "set" - a JSON-structured file that contains:

• A descriptive heading
• Key findings as bullet points
• Verified SQL queries for data acquisition
• Validated R code for analysis and visualization
• Metadata for organization and tracking

Sets are created sequentially with numerical prefixes (e.g., "01-overview.json", "02-detail.json") to maintain logical order in the final report.

AI Agents

The system integrates specialized AI agents, each with specific capabilities:

• Expert SQL generators for different blockchain ecosystems
• Analysis strategists for structuring investigations
• Data science experts for methodology validation

Each agent operates through a structured API interface and maintains its own context and expertise domain.

Workflow Architecture

1. Request Processing

When the system receives an analytical request, it:

• Creates a dedicated directory for the analysis
• Initializes a template structure
• Begins coordinating with appropriate agents

2. Set Creation Process

Each analysis set follows a strict verification workflow:

1. SQL Generation: Coordinates with expert agents to create appropriate queries
2. Data Validation: Verifies query execution and data structure
3. Analysis Creation: Develops and tests analytical code
4. Set Assembly: Combines verified components into a structured set
5. Format Verification: Confirms set structure and compatibility

3. Report Assembly

After all sets are created and verified, the system:

• Identifies all relevant JSON set files
• Orders them by numerical prefix
• Combines them into a cohesive report structure
• Generates the final HTML output

Core Functions

Agent Coordination

• Manages communication with specialized AI agents
• Ensures proper formatting of agent requests
• Validates agent responses before implementation

Code Execution

• Runs and verifies SQL queries in Snowflake
• Executes R code for analysis and visualization
• Maintains execution environment consistency

Set Management

• Creates numbered JSON set files
• Verifies set structure and content
• Manages set interdependencies

Report Generation

• Combines verified sets in proper order
• Assembles final report structure
• Handles HTML generation

Set Structure

{
  "sets": [{
    "heading": "Analysis Section",
    "bullets": ["Key finding 1", "Key finding 2"],
    "sql_chunk": {
      "object_name": "result_name",
      "query": "Verified SQL query"
    },
    "analysis_chunks": [{
      "object_name": "analysis_name",
      "code": "Verified R code"
    }]
  }],
  "metadata": {
    "title": "Analysis Title",
    "date": "2025-01-22",
    "directory": "analysis_dir"
  }
}