NTTP
Developer Tools
Query databases in plain English. 3-layer caching, multi-LLM support, auto error correction. SQL from natural language.
Key Features
- Natural language to SQL query translation
- 3-layer caching system (Exact, Semantic, LLM)
- Auto error correction for query refinement
- Multi-LLM support (Claude, GPT-4, Cohere, Mistral, Gemini)
- Multi-database support (PostgreSQL, MySQL, SQLite, SQL Server)
- Redis-based query persistence
- Full TypeScript type safety
- CLI and SDK interfaces for flexible integration
With NTTP (Natural Text Transfer Protocol), you can query databases in plain English instead of SQL.
Live Demo
Try NTTP with the interactive demo: nttp.appliedtensors.com
How It Works
NTTP caches at three layers:
- Exact Cache: Instant results for identical queries
- Semantic Cache: Matches similar questions with embeddings
- LLM Layer: Generates SQL when no cache hit exists
Key Capabilities
Multi-LLM Architecture: You can choose Claude, GPT-4, Cohere, Mistral, or Gemini for your needs and budget.
Auto Error Correction: When a query fails, NTTP automatically analyzes the error and refines the SQL. Manual debugging is not necessary.
Universal Database Support: NTTP works with PostgreSQL, MySQL, SQLite, and SQL Server through unified interfaces.
Production Ready: Queries survive restarts because Redis stores them. The whole codebase is typed.
Use Cases
- Rapid Prototyping: Query databases during development without SQL
- Data Analytics: Non-technical team members can explore data
- API Development: Build natural language interfaces for database access
- DevOps: Quick database inspections in plain language
GitHub Repository
Source code and detailed documentation: github.com/tylerbryy/nttp