DotaGPT

AI learning assistant for Dota 2 built with Gemini 2.5 Flash

Comment

Inspiration

As avid Dota 2 players, we noticed how overwhelming the game can be for newcomers - with so much things to learn making it the hardest MOBA game as of now. We wanted to create an AI-powered learning companion that breaks down these concepts into digestible, beginner-friendly content.

What it does

DotaGPT is an AI-powered chatbot designed specifically for new Dota 2 players. Using Gemini 1.5 Flash, it provides real-time answers to beginner questions about the 5 core positions (Carry, Mid, Offlaner, Roaming Support, Hard Support), role responsibilities, item recommendations, and basic gameplay mechanics. The chatbot serves as a personalized learning companion that helps newcomers understand complex Dota 2 concepts through interactive conversations.

How we built it

  • Frontend: Next.js with TypeScript for fast development and great developer experience
  • Styling: TailwindCSS for responsive, mobile-first design
  • Backend: Node.js API routes with MongoDB for storing position guides and chat responses
  • AI Chat: Integrated Gemini 1.5 Flash for real-time responses to beginner questions about Dota 2 positions and gameplay
  • Deployment: Vercel for seamless CI/CD integration

Challenges we ran into

  • Content curation: Distilling complex Dota 2 mechanics into truly beginner-friendly explanations without oversimplifying
  • AI prompt engineering: Fine-tuning Gemini 1.5 Flash prompts to provide accurate, beginner-focused Dota 2 advice within our domain
  • Token limitations: Limited Gemini API tokens affected development pace, requiring careful testing and optimization of chat features
  • Mobile optimization: Ensuring the chat interface works smoothly across devices, especially for gaming content
  • API rate limiting: Managing Gemini API calls efficiently during development and testing

Accomplishments that we're proud of

  • Created comprehensive position guides that genuinely help beginners understand their roles
  • Successfully integrated Gemini 1.5 Flash to provide intelligent, contextual responses about Dota 2 positions
  • Achieved excellent mobile responsiveness for both guides and chat interface
  • Delivered a polished, AI-powered educational tool within hackathon timeframe

What we learned

  • The importance of user-centered design when tackling complex gaming concepts
  • How to balance technical complexity with user simplicity
  • Effective content strategy for educational applications
  • Modern web development best practices with Next.js and MongoDB integration

What's next for DotaGPT

  • Enhanced AI context: Train custom models or fine-tune prompts with more specific Dota 2 knowledge
  • Hero recommendations: AI-powered hero suggestions based on position and beginner skill level
  • Interactive tutorials: Hands-on learning modules with AI guidance
  • Advanced conversations: Multi-turn chat sessions with memory for personalized learning paths
  • Community features: AI-moderated beginner forums and Q&A sections

Built With

Share this project:

Updates