Project Overview
A ChatGPT-driven interface that converts natural-language questions into SQL, lets users
review/edit queries, executes them on custom databases, and returns results in plain language
Challenges
- Accurate SQL generation from free-form language: Translating varied user phrasing
into syntactically correct, efficient SQL (including JOINs, aggregates, filters).
- Handling complex, multi-table queries: Ensuring generated SQL respects relationships,
performance constraints, and returns the intended results without over-fetching.
- Safe user editing and execution: Allowing users to modify generated SQL while
preventing accidental destructive queries or syntax/runtime errors.
- Balancing automation vs. control: Providing a smooth, fast experience for casual users
while offering experts the ability to review and refine queries.
- Extending functionality beyond SQL: Planning to support documents and CSVs
introduced formats and retrieval challenges (indexing, schema inference, large-file
handling).
Solutions
- Accurate SQL generation from free-form language: Translating varied user phrasing into syntactically correct, efficient SQL (including JOINs, aggregates, filters).
- Handling complex, multi-table queries: Ensuring generated SQL respects relationships,
performance constraints, and returns the intended results without over-fetching.
- Safe user editing and execution: Allowing users to modify generated SQL while
preventing accidental destructive queries or syntax/runtime errors.
- Balancing automation vs. control: Providing a smooth, fast experience for casual users
while offering experts the ability to review and refine queries.
- Extending functionality beyond SQL: Planning to support documents and CSVs
introduced formats and retrieval challenges (indexing, schema inference, large-file
handling).
Technology Used
- Programming Language: Python
- AI / NLP: ChatGPT API (direct integration; LangChain used during prototyping)
- Backend: Flask (API layer, validation, execution logic)
- Frontend: React (query input, SQL preview/editor, results display)
- Database: MySQL (customer-provided)
- SQL Generation & Safety: Custom logic for SQL construction, validation, and
execution controls (timeouts, read-only defaults, sanitization)
- Deployment / Scaling: Docker (containerized services)
- Additional / Planned: Document & CSV query adapters (schema inference,
chunking/embedding layers planned)
