Documentation

Introduction

Veronica is a sophisticated AI intelligence system combining mentorship capabilities with research-grade reasoning and analysis. The system operates through a multi-agent architecture that coordinates specialized agents for comprehensive research tasks.

The documentation is organized into sections covering setup, usage, API reference, and system architecture. Whether you're using the web interface or integrating programmatically, you'll find guidance here.

Quick Start

Get started with Veronica in minutes. The research interface provides direct access to multi-agent research capabilities without requiring any setup.

Basic Research Query

Simply enter your research query and select your preferred search provider:

Configuration

For advanced usage and CLI access, configure your environment with API keys for preferred search providers.

# Set your API keys as environment variables
export MOONSHOT_API_KEY="your_moonshot_key_here"
export EXA_API_KEY="your_exa_key_here"

# Optional: Set default provider
export VERONICA_DEFAULT_PROVIDER="moonshot"

Research Interface

The web interface provides real-time visibility into the multi-agent research process. Watch as agents progress through four stages: Search, Analyze, Verify, and Synthesize.

Interface Components

• Command Palette: Enter queries and configure search options
• Metric Cards: Real-time stats on sources, claims, and confidence
• Progress Cards: Visual tracking of each research stage
• Activity Log: Timestamped events during research process
• Source Panel: Filtered list of discovered sources with credibility ratings

Command Line Interface

The CLI provides full access to Veronica's capabilities from your terminal. Ideal for automation, scripting, and integration into workflows.

# Basic research query
veronica research "climate change solutions"

# Specify provider and category
veronica research "AI safety" --provider exa --category research

# Export to markdown
veronica research "quantum entanglement" --output report.md

# JSON output for automation
veronica research "neural networks" --format json --output data.json

Common Options

• --provider: Select search provider (moonshot, exa, auto, hybrid)
• --category: Filter by category (all, research, news, web)
• --output: Specify output file path
• --format: Output format (text, markdown, json)
• --verbose: Enable detailed progress output
• --sources: Maximum number of sources (default: 20)

Understanding Outputs

Veronica produces structured research outputs following the ResearchParadigm format. Each output includes verified claims, source references, and confidence metrics.

Output Structure

• Summary: High-level overview of findings
• Key Findings: Verified claims with confidence scores
• Sources: Cited references with credibility ratings
• Metadata: Research parameters and processing stats

{
  "query": "artificial general intelligence timeline",
  "summary": "Expert estimates for AGI range from 2030 to 2050...",
  "confidence": 0.87,
  "findings": [
    {
      "claim": "Median expert estimate places AGI around 2040",
      "confidence": 0.92,
      "sources": ["source-001", "source-003", "source-007"]
    }
  ],
  "sources": [
    {
      "id": "source-001",
      "title": "AI Progress Forecasting",
      "url": "https://example.com/paper",
      "credibility": 0.95
    }
  ],
  "metadata": {
    "providers_used": ["moonshot", "exa"],
    "processing_time": 42.3,
    "sources_analyzed": 23
  }
}

API Overview

The Veronica API provides programmatic access to research capabilities. RESTful endpoints support both synchronous and asynchronous operations.

Authentication

Include your API key in the request header for authenticated requests:

curl -X POST https://api.veronica.ee/v1/research \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "machine learning interpretability"}'

API Endpoints

Webhooks

Configure webhooks to receive notifications when research completes. Webhooks are sent as POST requests with JSON payloads.

Webhook Payload

{
  "event": "research.complete",
  "timestamp": "2025-02-18T12:34:56Z",
  "data": {
    "id": "req_abc123",
    "query": "original query text",
    "status": "complete",
    "confidence": 0.89,
    "sources_count": 15
  }
}

Architecture

Veronica operates on a multi-agent architecture where specialized agents coordinate to handle different aspects of the research process. Each agent has specific capabilities and communicates through a central coordination layer.

Agent Types

• Search Agent: Discovers relevant sources across configured providers
• Analysis Agent: Extracts and processes information from sources
• Verification Agent: Cross-references claims and assesses credibility
• Synthesis Agent: Combines findings into coherent output
• Coordinator Agent: Orchestrates workflow and manages state

Providers

Veronica integrates with multiple search providers to offer comprehensive coverage of different content types and sources.

Available Providers

Troubleshooting

Common issues and their solutions.

Research Takes Too Long

Reduce the --sources parameter or use a single provider instead of hybrid mode.

Low Confidence Scores

Try reformulating the query to be more specific, or switch to a provider specialized in your topic area.

API Authentication Errors

Verify your API key is correctly set in the Authorization header. Keys must be prefixed with "Bearer ".