
RivalRadar is an autonomous AI research agent built with Gemini 2.5 Flash Lite and Bright Data APIs. It works like a human analyst — but 100x faster. You type a company name, and the agent autonomously: - Searches Google in real-time for top competitors - Scrapes actual pricing pages, news articles, and LinkedIn job listings - Reasons over collected data using Gemini 2.5 Flash Lite - Generates a structured competitive intelligence report with visual charts Key Features: - Live Google Search via Bright Data SERP API - Deep web scraping via Bright Data Scraping Browser - Real-time streaming via Server-Sent Events - Visual dashboard with Chart.js — market share, funding comparison - Strengths & weaknesses auto-extraction - Search history with local storage - Export report as HTML Built on MCP (Model Context Protocol) architecture — tools are plug-and-play, making it easy to add new data sources like LinkedIn API, Crunchbase, or Slack notifications. Tech Stack: Python, Flask, Gemini 2.5 Flash Lite, Bright Data, Chart.js, BeautifulSoup4
31 May 2026

SynapseOS is a multi-agent AI debate system built on AMD MI300X GPU infrastructure. When a user submits any business idea or problem, five specialized AI agents independently analyze it from their unique professional perspectives. The PM Agent breaks down project phases and timelines. The Developer Agent recommends tech stack and architecture. The Critic Agent identifies market risks and flaws. The Finance Agent projects costs and revenue models. The Security Agent flags vulnerabilities and compliance requirements. Each agent runs Qwen2.5-0.5B-Instruct via vLLM on AMD Developer Cloud, providing 150+ word detailed responses. The system concludes with a final GO/CONDITIONAL GO/NO-GO verdict from the PM Agent, plus a voice summary generated via gTTS. Built with Gradio UI, deployed on HuggingFace Spaces, and powered by AMD ROCm stack — SynapseOS demonstrates practical multi-agent AI orchestration for real-world decision support.
10 May 2026