<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Rushi Patil</title>
    <description>The latest articles on DEV Community by Rushi Patil (@rushi_patil_0b5da6dc3e9cf).</description>
    <link>https://dev.to/rushi_patil_0b5da6dc3e9cf</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3936283%2Fdd52f405-f189-4ca0-b9b3-8cfc07179cdb.png</url>
      <title>DEV Community: Rushi Patil</title>
      <link>https://dev.to/rushi_patil_0b5da6dc3e9cf</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/rushi_patil_0b5da6dc3e9cf"/>
    <language>en</language>
    <item>
      <title>Building "Cric-Oracle Pro": A Live Multi-Agent IPL Strategist using Gemini 3.1 Flash-Lite</title>
      <dc:creator>Rushi Patil</dc:creator>
      <pubDate>Sun, 17 May 2026 12:56:14 +0000</pubDate>
      <link>https://dev.to/rushi_patil_0b5da6dc3e9cf/building-cric-oracle-pro-a-live-multi-agent-ipl-strategist-using-gemini-31-flash-lite-2bgc</link>
      <guid>https://dev.to/rushi_patil_0b5da6dc3e9cf/building-cric-oracle-pro-a-live-multi-agent-ipl-strategist-using-gemini-31-flash-lite-2bgc</guid>
      <description>&lt;p&gt;Hey everyone! 👋 I just wrapped up an intense 3-hour solo sprint for the latest GDG Pune hackathon. The challenge was to build an agentic AI system that acts as a virtual IPL captain.&lt;/p&gt;

&lt;p&gt;Instead of building a generic chatbot that just spits out statistics, I wanted to build something that models the human element of the game. Enter Cric-Oracle Pro—a live, multi-agent war room.&lt;/p&gt;

&lt;p&gt;🧠 The Architecture: Going Beyond the Math&lt;br&gt;
Cricket isn't just played on the pitch; it's played in the mind. To replicate a real IPL captain's thought process, I orchestrated a multi-turn reasoning loop using three distinct AI agents powered by Gemini 3.1 Flash-Lite:&lt;/p&gt;

&lt;p&gt;📊 The Data Analyst: Handles the raw math, required run rates, and historical batter-vs.-bowler match-ups.&lt;/p&gt;

&lt;p&gt;🔥 The Momentum Architect: The secret sauce of the app. This agent ignores stats completely and analyzes the "Choke Factor," the fear metric, and the psychological momentum of the crowd.&lt;/p&gt;

&lt;p&gt;🏆 The Captain: Synthesizes the quantitative data and the psychological audit to issue one definitive, bold tactical mandate for the next over.&lt;/p&gt;

&lt;p&gt;🛠️ The Tech Stack&lt;br&gt;
Since the rules strictly required the Google ecosystem, I built the entire backend natively:&lt;/p&gt;

&lt;p&gt;AI Orchestration: gemini-3.1-flash-lite via the modern google-genai Python SDK.&lt;/p&gt;

&lt;p&gt;Backend: FastAPI for asynchronous, high-speed agent routing.&lt;/p&gt;

&lt;p&gt;Frontend: Pure HTML &amp;amp; Tailwind CSS. I ditched basic UI frameworks to build a custom, glassmorphism "cyber-broadcast" layout that looks like a premium enterprise sports tool.&lt;/p&gt;

&lt;p&gt;Live Radar: A custom Python scraping tool that feeds real-time live match data directly into the Gemini context window.&lt;/p&gt;

&lt;p&gt;🚀 The Result&lt;br&gt;
The system doesn't just say "Bowl a spinner." It says, "The required rate has breached 12. The batter is desperate, and the crowd is silent. Bring mid-off inside the circle to cut off the single, suffocate the strike, and force a panicked shot over the top."&lt;/p&gt;

&lt;p&gt;It was an incredible experience pushing the boundaries of Gemini's multi-agent capabilities under a strict time limit. Huge shoutout to the community for hosting this!&lt;/p&gt;

&lt;h1&gt;
  
  
  gdgcloudpune #gdgpune #gemini #artificialintelligence #python #hackathon
&lt;/h1&gt;

</description>
      <category>agents</category>
      <category>ai</category>
      <category>gemini</category>
      <category>showdev</category>
    </item>
  </channel>
</rss>
