Web Application
T-shooting Heuristics
1. Frontend:
- Next.js as the main framework (React + server components)
- Tailwind CSS for styling
- Vercel for hosting
- PWA functionality for mobile device support
2. Knowledge Base:
- Supabase for storing structured data
- MDX files for content storage (examples, instructions)
- Vector embeddings for semantic search
3. AI Integration:
- Anthropic API for working with Claude
- Langchain for prompt management
- Pinecone for vector search in the knowledge base
Basic Functionality of v1.0:
1. Main Interface:
- Chat interface for interaction with the AI assistant
- Sidebar with quick access to:
- Method database
- Example collection
- Checklists
- Personal notes
2. Core Features:
- Task formulation with the help of AI
- Selection of relevant methods and examples
- Step-by-step hints for method application
- Solution history saving
- Export of results
3. Workflow Mechanics:
a) User Entry:
- Problem description in free form
- AI helps structure the task
- The system suggests relevant tools
b) Problem-Solving Process:
- Step-by-step method application with hints
- Access to examples of similar solutions
- Intermediate result quality checks
- Idea generation with the help of AI
c) Documentation:
- Automatic process saving
- Report generation
- Creation of a personal experience database
4. Integration with Claude:
- Pre-prepared prompts for various scenarios
- Call function system for:
- Determining the task type
- Method selection
- Formulation checks
- Analogy search
- Idea generation
- Contextual learning based on feedback
Implementation Plan:
1. Week 1-2:
- Deploy basic infrastructure
- Create an interface prototype
- Set up integration with the Anthropic API
2. Week 3-4:
- Populate the knowledge base
- Develop main prompts
- Create the navigation system
3. Week 5-6:
- Implement core functionality
- Test performance
- Collect feedback
4. Week 7-8:
- Optimize performance
- Improve UX
- Prepare for launch