EDAUtils AI System

Platform Overview

The EDAUtils AI System is the central intelligence platform that orchestrates all EDA tools through intelligent routing and continuous learning. It provides a unified interface for semiconductor engineers to access AI-powered analysis, code generation, and verification tools.

Built on a distributed client-server architecture, the platform enables multi-user concurrent access while maintaining data security through local processing. The system continuously learns from user interactions to improve tool selection and response quality.

Platform Capabilities

Intelligent Tool Selection with LLM-powered routing
Multi-LLM Backend (Ollama, Claude, Gemini, OpenAI)
Continuous Learning (RLHF, Active Learning, Curriculum)
Multi-User Concurrent Access
Workspace-Aware File Management
Extensible Plugin Architecture

How It Works

User Input

Engineers submit queries through the local agent:

Analyze CDC violations in my design
Generate testbench for this module
Extract protocol from Verilog file
Create formal properties from spec

System Output

Platform intelligently routes to appropriate tool and returns:

Detailed analysis with root causes
Generated code or specifications
Automated fixes or recommendations
Learning insights for improvement

Technical Architecture

Local Agent: Runs on engineer's workstation, manages workspace scanning, file metadata extraction, and server communication. Provides CLI interface and feedback collection.

Central Server: Flask-based API server with concurrent orchestrator, tool registry, and intelligent routing. Handles multiple users simultaneously with rate limiting and session management.

LLM Manager: Abstracts multiple LLM backends (Ollama, Claude, Gemini, OpenAI) with connection pooling and load balancing. Provides unified interface for all tools.

Learning System: Implements RLHF, active learning, and curriculum learning. Continuously improves tool selection and response quality based on user feedback.

Library of EDA Utilities