This is a submission for the GitHub Copilot CLI Challenge
What I Built
Vish AI is a mental health companion that's actually available when you need it. No waitlists, no $200/hour sessions, no judgment.
332 million people worldwide deal with depression. Suicide is the third leading cause of death for 15-29 year olds. Traditional therapy has month-long waitlists and costs $100-250 per session. Crisis hotlines only work during emergencies. Mental health apps? 70-75% abandoned within 100 days.
Vish AI learns from your journals, medical records, and personal history to provide context-aware support. Text or voice, whenever you need it. Built using Azure OpenAI, LangChain for conversation memory, and RAG for depth using mental health and spirituality resources.
Important: This is an LLM-powered tool. It makes mistakes. If you're in crisis, seek professional help immediately.
Demo
Try it out: VishAI
My Experience with GitHub Copilot CLI
GitHub Copilot CLI transformed how I built this project. Instead of switching between browser tabs for documentation or spending time on boilerplate code, I stayed in the terminal and got work done.
Where Copilot CLI made the biggest impact:
- Setting up Azure integrations: Instead of manually reading Azure OpenAI docs, I described what I needed and Copilot CLI generated the configuration code and deployment scripts.
- LangChain conversation memory: Implementing conversation context required understanding LangChain's memory management. Copilot CLI helped me write the correct chain configurations and memory buffer setup without trial-and-error debugging.
- RAG implementation: Loading mental health resources into the vector database required specific data preprocessing. Copilot CLI suggested optimal chunking strategies and embedding generation code.
- Debugging Azure deployment issues: When deployment failed, I described the error to Copilot CLI and it suggested fixes for authentication issues and resource configuration problems.
The development speed difference:
- Before Copilot CLI: Googling syntax, reading docs, copying examples, adapting to my code → 20-30 minutes per feature
- With Copilot CLI: Describe what I need, review suggestion, implement → 5-10 minutes per feature
Top comments (1)
Let me know what you think and how I can improve it :)
Don't forget to like 💖