Ashoka

An AI-powered governance advisor for Internet Computer Protocol realms and DAOs. Ashoka provides intelligent responses to governance questions using multiple AI personas, powered by LLMs and retrieval-augmented generation (RAG).

Features

• Multi-Persona AI Governance: Choose from specialized personas (advisor, facilitator, analyst)
• Realm Status Integration: Context-aware responses using real-time realm data
• CLI & HTTP API: Flexible interfaces for different use cases
• RunPod Cloud Deployment: Scalable GPU-powered infrastructure
• Conversation History: Track governance discussions and decisions

Quick Start

Using CLI (Recommended)

# Ask a governance question

# Use specific persona for strategic advice
./pod_manager.py main ask -q "Should we approve this treasury proposal?" -p advisor

# Ask complex questions from file with realm context
./pod_manager.py main ask -qf complex_proposal.txt -p facilitator --realm-status-file realm_data.json

# List available personas
./pod_manager.py main personas

# Check API health
./pod_manager.py main health

# Deploy new pod with cheapest available GPU
./pod_manager.py main deploy

# Start existing pod (deploy new if needed)
./pod_manager.py main start --deploy-new-if-needed --verbose
# Check pod status
./pod_manager.py main status

Pod Operations

# Stop pod
./pod_manager.py main stop

# Restart pod
./pod_manager.py main restart

# Terminate pod (delete)
./pod_manager.py main terminate

API Commands

Ask Questions

# Simple question
./pod_manager.py main ask -q "What is quadratic voting?"

# Question from file (for long/complex questions)
./pod_manager.py main ask -qf governance_proposal.txt

# With specific persona
./pod_manager.py main ask -q "Analyze this budget proposal" -p advisor

# With realm context data
./pod_manager.py main ask -q "Should we proceed?" --realm-status-file current_state.json

Persona Management

# List all available personas
./pod_manager.py main personas

# Get details for specific persona
./pod_manager.py main persona -p ashoka

# Available personas: ashoka (default), advisor, facilitator

Realm Status

# Get status for all tracked realms
./pod_manager.py main realm-status

# Get status for specific realm
./pod_manager.py main realm-status -r realm_principal_id

Health Check

# Check API health and connectivity
./pod_manager.py main health

Architecture Overview

• Pod Manager: RunPod cloud deployment and management
• API Service: Flask HTTP API with CORS support
• Persona System: Multiple specialized AI governance advisors
• Database: PostgreSQL for conversation history and realm status
• LLM Integration: Ollama for local LLM inference
• RAG System: ChromaDB for document embeddings and retrieval

Configuration

Set environment variables:

export RUNPOD_API_KEY="your_runpod_api_key"
export ASHOKA_DEFAULT_MODEL="llama3.2:1b"

File Structure

• pod_manager.py - Main CLI tool for pod management and API interaction
• api.py - HTTP API service
• persona_manager.py - Multi-persona system
• database/ - Database client and schema
• prompts/personas/ - AI persona definitions
• tests/ - Test cases and scenarios