Gregify

• 

  Our landing page

• 

  Our backend Agents

• 

  Gregify in action

Inspiration

The world of AI prompting has become increasingly complex, yet the tools to help users craft effective prompts remain surprisingly basic. We noticed that while prompt engineering was becoming a crucial skill, there wasn't a seamless, intelligent solution integrated directly into users' browsing experience. The light bulb moment came when we realized we could combine the power of RAG (Retrieval-Augmented Generation) and multi-agent systems (MAS) to generate prompts that the users can use to get more transparent, more precise answers to their questions.

What it does

Gregify is a Chrome extension that transforms basic prompts into optimized, context-aware queries. Through a simple interface, users can:

• Input their initial prompt
• Select specific AI models they're targeting
• Choose specialized agents for different optimization strategies
• Receive enhanced prompts that leverage best practices and proven patterns
• Benefit from multi-agent system analysis that approaches prompt optimization from multiple perspectives simultaneously

How we built it

We developed Gregify using a modern tech stack that combines frontend and backend innovations:

• Chrome Extension: Built with React and TypeScript for a responsive interface
• RAG System: Implemented using a vector database to store and retrieve relevant prompt engineering patterns and examples
• Multi-Agent System: Developed multiple specialized agents that collaborate to analyze and enhance prompts from different angles (clarity, specificity, context-awareness, etc.)
• Backend API: Created using FastAPI and a Webhook to handle communication between the extension and our MAS/RAG systems

Challenges we ran into

• Implementing real-time prompt analysis while maintaining extension performance
• Coordinating multiple agents to provide coherent, unified prompt improvements
• Building an effective RAG system with high-quality prompt engineering examples
• Managing the complexity of context-aware prompt enhancement without overwhelming users
• Optimizing the extension to work smoothly across different websites and platforms

Accomplishments that we're proud of

• Created a novel approach to prompt enhancement using RAG and multi-agent systems
• Built a user-friendly interface that makes advanced prompt engineering accessible
• Successfully implemented real-time prompt optimization in a browser extension
• Developed a scalable architecture that can accommodate future improvements
• Completed a working prototype in a short timeframe for the hackathon
• Created a landing page for Gregify

What we learned

• The intricacies of building Chrome extensions with complex backend systems
• How to effectively combine RAG and multi-agent systems for practical applications
• The importance of user experience in technical tools
• Techniques for optimizing performance in browser-based AI applications
• The value of rapid prototyping and iterative development

What's next for Gregify

• Expanding the RAG database with more specialized prompt engineering patterns
• Adding support for more AI models and platforms
• Implementing user feedback loops to improve prompt enhancement accuracy
• Developing customizable agent configurations for different use cases
• Creating a community feature for sharing and rating enhanced prompts
• Adding analytics to help users understand how their prompts are being improved

Built With

• googledriveapi
• microsoftautogen
• n8n
• openaiapi
• openaigpt4ominimodel
• openaitextembeddingmodel
• perplexityapi
• react
• shadcn
• supabase
• typescript