Natural Language SQL

Overview

Natural Language SQL is a self-hosted engine for turning database questions written in everyday language into executable SQL. It is best suited to developers and teams who want local AI and the chance to inspect queries before they run. Its privacy-conscious defaults and open-source license are appealing, but a large local model and write-enabled queries call for careful setup.

Key features

Hybrid schema retrieval combines BM25 search with sentence-transformer vectors to find relevant database context for each question. That can help users ask about an existing schema without assembling context by hand, while the reported 52.32% execution accuracy on Spider dev underscores why generated SQL should not be treated as infallible.

Users can edit generated SQL before execution, and streamed progress makes the query workflow visible. Explanations, sorting, 50-row pagination, CSV and JSON downloads, and tab-separated copying support inspection and handoff. SELECT queries get a default LIMIT; explicit result counts are capped at 1,000 rows, and execution times out after 30 seconds by default.

The safeguards are not a substitute for access control: INSERT, UPDATE, and DELETE are allowed by default, with warning banners. The project recommends connecting with a read-only database user if writes must be blocked.

Ollama runs locally by default, and the project says questions, schema, and generated SQL stay on the user's machine with that provider. There is no telemetry; schema metadata, query results, and query history remain in session memory rather than being written to disk by the application. Choosing OpenAI, Gemini, or Groq sends the question, relevant schema portion, and SQL context to that provider, but not database passwords or result rows. Database passwords and cloud API keys are Fernet-encrypted at rest, redacted from logs, and masked in API responses.

Pricing

The Natural Language SQL Engine costs nothing. It is a GPL-3.0 open-source project for local installation, with no paid tier or usage quota described. That makes it a fit for users willing to run and maintain the software themselves; those seeking a managed hosted plan should consider alternatives.

Platforms

Natural Language SQL is offered for API, Linux, macOS, self-hosted, and web use. Its recommended Docker deployment needs Docker 20.10 or later with Compose v2 and about 10 GB of free disk space; the first run downloads an approximately 5 GB model. That footprint is a meaningful cost in storage and setup effort, especially on smaller machines.

Who it's for

This is a sensible choice for developers who work with PostgreSQL or MySQL, want to query multiple databases of the same type in parallel, and can manage a local deployment. Local Ollama is particularly suited to users who prefer database context to stay on their machine. It is less appropriate when writes must be prevented without a read-only account, or when the deployment requirements are impractical.

Pros and cons

• Pros: Local Ollama keeps questions, schema, and SQL on the user's machine, and the project says it collects no telemetry.
• Pros: Editable SQL, explanations, streamed progress, and practical export options support review rather than blind execution.
• Pros: Free, open-source local installation connects to PostgreSQL and MySQL, including multiple same-type databases in parallel.
• Cons: Docker deployment calls for about 10 GB of free disk space, with an approximately 5 GB model download on first run.
• Cons: Write queries are enabled by default, so safe use against protected data depends on database permissions.
• Cons: The reported Spider dev execution accuracy is 52.32%, making human review important.

Alternatives

AI SQL Generators is the broader category to browse when comparing tools.

Choose Outerbase AI if a freemium option with a free tier capped at five users, one base, and 10 EZQL queries a month suits better than a local open-source engine.

NatureQuery offers a free tier with 50 queries per month, one database connection, exports, and 30-day query history.

SQL Mocker has a free Explore tier with 50 AI queries total, one user, and five saved projects.

Wren AI is an option for individual developers seeking an open-source context engine through CLI and MCP without a UI.

QueryPlane offers a free tier for up to three apps, one database connection, and 50 AI prompts per month.

mnemiq is another free, open-source engine designed to run in your own environment.

Text2SQL.ai is a paid option with a free trial and API access listed for its Pro Plan.

Vanna AI may suit readers looking for a freemium product with a paid Explorer plan that includes 20 questions per day, API access, and same-day email support.

Verdict

Choose Natural Language SQL if you can self-host, want local AI, and value the ability to inspect and edit SQL before it runs. Its strongest case is keeping database context on your machine without a software charge; look elsewhere if you need managed deployment or cannot rely on a read-only database account to prevent writes.