AI Agent - June 20, 2025 - Ask Racha - PLDG 3

[Dhruv-Varshney-developer]

AI Agent - June 20, 2025 - Ask Racha - PLDG 3

---

Contributors

@Dhruv-Varshney-developer

---

Accomplished Tasks

• Initialized full-stack Ask Racha project from scratch
• Implemented basic RAG system
• Integrated Gemini API + LlamaIndex for accurate RAG
• Built and deployed backend Flask app on Render
• Created initial frontend in raw HTML + Flask integration
• Replaced raw HTML with a modern Next.js frontend
• Refactored API URL configs from hardcoded to .env-based
• Updated CORS allowed origins for production deployment
• Added helpful project docs and env.example for local use
• Frontend deployed on Vercel, backend on Render
• Multiple RAG iterations and final stable release shipped

---

Update

• Started and shipped Ask Racha from zero to production.
• Implemented a proper RAG pipeline using Gemini API (free, stable) + LlamaIndex.
• Replaced OpenAI because of rate limits — Gemini performed smoothly.
• First few iterations included basic RAG and frontend with Flask.
• Revamped into a clean Next.js frontend for sleek UI and better UX.
• Configured .env for flexible dev and prod environments.
• Cleaned up repo, added env.example, updated CORS origins for cross-platform compatibility.
• App now fully live and stable.
  • Frontend: https://ask-racha.vercel.app
  • Backend: https://ask-racha.onrender.com

What went well:

• End-to-end execution: backend, frontend, deployment
• Gemini API worked like a charm for RAG tasks
• Frontend revamp made the tool highly usable and demo-ready

What was difficult:

• Troubleshooting CORS and deployment configs across platforms
• OpenAI rate-limiting issues early on

---

Demo Videos

AskRacha.Demo.mov

---

Questions

None for now — everything moving smoothly. Feedback welcome on the project.

---

Next Tasks

• Train Ask Racha on the entire Storacha documentation starting from https://docs.storacha.network/
• Current limitation: Only the main page is being crawled. The bot does not follow or ingest sublinks (e.g., https://docs.storacha.network/w3up-client/)
• Next task is to implement recursive crawling to fetch all nested documentation pages and feed them into the RAG pipeline
• Once all subpages are ingested, improve response depth and coverage across the entire Storacha knowledge base

Links:

1. https://github.com/Dhruv-Varshney-developer/ask-racha
2. App live at: https://ask-racha.vercel.app