MVPApril 15, 2026

Vyuha-Mind

An autonomous strategy engine that deploys LLM-powered agents to stress-test your product against real market chaos — before you ship.

PythonFastAPIReactVite

Source

Yashwanth Kamireddi · AI Engineer · Full-Stack Developer

Overview

Vyuha-Mind is a multi-agent product strategy simulator. You feed it a product spec (name, features, price, target market) and a market scenario (region, geopolitical context); it deploys a team of LLM-powered agents to simulate the launch and hands you back a phased GTM playbook — grounded in real market signal, not vibes.

How it works

A multi-agent orchestrator runs the simulation in parallel: discovery agents pull live market context, strategy agents propose moves, risk agents stress-test them under crisis scenarios, and a synthesis agent writes the final recommendation. Each agent has a narrow role, a tool budget, and explicit handoff contracts.

AI stack

Live LLM mode — Google ADK on Gemini 2.0 Flash with Grok API as a secondary signal source
Free fallback — when API keys aren't set, agents pull from Google News RSS, Reddit, Wikipedia pageviews, and Google Trends to simulate market context for free
Backend — FastAPI + Uvicorn, Python orchestrator (agents.py, orchestrator.py, adk_runtime.py)
Frontend — React 19 + Vite, dashboard with simulation telemetry, scenario presets, Monte Carlo run history

Highlights

50+ autonomous agents per simulation run
Under 15 seconds end-to-end deliberation cycle on Gemini 2.0 Flash
1000+ Monte Carlo runs supported per scenario
6 starter scenario templates (price war, supply shock, geopolitical sanctions, etc.)

Gallery

Back to all projects