Back to Projects
Live

AutoArchitect

Multi-Agent Architecture Design Tool

Python Flask Google ADK Gemini 2.0 Flash Mermaid.js

Motivation

Exploring how agentic AI can transform the way we design software systems. This project demonstrates how coordinated, task-driven agents can produce high-level technical documentation from a simple natural language description.

What it Generates

Structured Requirements Analysis

Extracts functional and non-functional requirements from project descriptions.

Architecture Patterns

Recommends architecture patterns and technology selections.

C4-Style Diagrams

System Context, Container, Sequence, and Data Model diagrams via Mermaid.js.

Cloud Cost Estimates

Estimates AWS/GCP/Azure costs for the proposed architecture.

Architecture Review

Validates architectural decisions and suggests improvements.

Agent Pipeline

01

Orchestrator Agent — Manages the full workflow, coordinating all downstream agents.

02

Requirements Agent — Extracts functional and non-functional requirements from the description.

03

Pattern Agent — Selects architecture patterns and technology recommendations.

04

Diagram Agent — Produces Mermaid.js diagrams for all C4 levels.

05

Review Agent — Validates architectural decisions against best practices.

06

Costing Agent — Estimates cloud infrastructure costs across providers.

Tech Stack

Python Flask Google ADK Gemini 2.0 Flash Mermaid.js Vanilla JS Render