AI2SQL

Plain English to SQL that runs on your real schema

Intermediate API
Screenshot of AI2SQL, Plain English to SQL that runs on your real schema

What is AI2SQL?

AI2SQL is a developer tool. AI SQL assistant that writes, explains, fixes, optimizes and formats SQL against your connected schema, then runs it read-only. Ships as a web app, a Windows and macOS desktop connector, an MCP server for Claude, ChatGPT and Cursor, and a governed gateway for team agents.

AI2SQL is an AI SQL assistant that converts a plain English question into a SQL query written against your own connected database, not a generic example schema. You point it at a Postgres, MySQL, SQL Server, Oracle, Snowflake, BigQuery, Redshift, MongoDB, SQLite or Firebird instance using read-only credentials, and it builds a semantic layer from your tables, joins and business terms. Ask for the result you want and it returns SQL in that database's dialect with your actual table and column names, which you can execute read-only to see the rows. The product covers more than generation. Explain produces a clause by clause walkthrough of a query you inherited. Fix repairs syntax errors, dialect mismatches and wrong column names while describing what broke. Optimize rewrites a slow query and explains the reasoning, such as a join that should be a filter or a missing index. Format normalizes casing and indentation so output is readable before it reaches a pull request. Every generated statement passes through sqlGuard, which classifies it and blocks INSERT, UPDATE, DELETE, DROP, other DDL, multi statement input and functions that read server files. Credentials are encrypted and each query is logged and metered. There are three delivery routes. The desktop Connector for Windows and macOS runs beside your database so credentials stay on your machine, which is also how you reach a host that only listens on localhost or an internal hostname. The MCP endpoint lets Claude, ChatGPT in developer mode, Cursor or your own agent call run_query from a chat window. The gateway gives a team one governed read-only endpoint with per agent keys. A REST API is available alongside MCP. Developers and analysts are the core audience, and learners benefit from seeing their own question mapped to SELECT, JOIN and GROUP BY. The honest comparison is ChatGPT: for a single query on a schema you can paste into a chat box, the two are close. AI2SQL pulls ahead when the work needs a live database, the right dialect, execution and a write block. The Free plan is limited to 3 queries a day against a demo database, and connecting your own database requires a paid tier.

How do you use AI2SQL?

  1. 1Create an account and open the builder
    Sign up from the homepage and land in the query builder, where a demo database with sample tables is available on the Free plan.
    AI2SQL — Create an account and open the builder
  2. 2Connect a read-only database
    For your own data, add a Postgres, MySQL, SQL Server, Snowflake, BigQuery or other supported connection using read-only credentials. This needs a paid plan.
  3. 3Describe the result you want
    Type the question in plain English. AI2SQL writes the SQL in the dialect of the selected database using your real table and column names.
  4. 4Run it read-only and inspect the rows
    Execute the statement against the live read-only connection and check the returned rows inside the builder before trusting the answer.
  5. 5Ask your database from your existing AI client
    Connect AI2SQL as an MCP server in Claude, ChatGPT or Cursor, then call run_query from the chat window with a scoped agent key.
    AI2SQL — Ask your database from your existing AI client
  6. 6Run the connector next to a private database
    Install the Windows or macOS desktop app when the database only listens on localhost or an internal hostname, keeping credentials on your machine.
    AI2SQL — Run the connector next to a private database

Pros and cons

Pros

  • Works from your live schema, so table names, columns and joins match your actual database rather than a textbook exampleAI
  • Covers the full query lifecycle: generate, explain, fix, optimize, format and execute read-only in one placeAI
  • sqlGuard blocks writes, DDL, multi statement input and server file functions before anything reaches the databaseAI
  • Connects to ten database engines including Snowflake, BigQuery, Redshift, MongoDB and FirebirdAI
  • MCP and REST endpoints let existing AI agents query the database with per agent keys and logged callsAI

Cons

  • Free plan is capped at 3 queries a day and only works on a demo database; connecting your own data requires a paid planAI
  • Query generation still depends on a connection to the vendor's API, so there is no fully offline mode even with the desktop appAI
  • The desktop connector, the only route to localhost or internal hosts, is restricted to the Pro tierAI
  • Generated SQL still needs human review for joins and filters, and the vendor itself recommends checking the execution plan for anything run oftenAI
  • Starting the 7 day trial requires a card, which adds friction for anyone who wants to test before committingAI

How much does AI2SQL cost?

Free trial

Every plan starts with a 7 day free trial. No charge is made on the first day, but a card is required to begin, and the trial can be cancelled in one click from the account before it ends.

Learn more

Pricing

Free plan includes 3 queries a day on a demo database. Paid tiers are Start at $9 per month, Pro at $19 per month and Team at $39 per month. Connecting your own database requires Pro or Team, and the Windows and macOS desktop app is a Pro feature.

Learn more

Integrations

Ten database connectors: PostgreSQL, MySQL, SQL Server, Oracle, Snowflake, BigQuery, Amazon Redshift, MongoDB, SQLite and Firebird. MCP works with Claude, ChatGPT in developer mode, Cursor and custom agents, and a REST API is also available.

Learn more

Features

Generate SQL from plain English against your connected schema, then explain, fix, optimize and format any query, generated or hand written. Execute read-only and read the rows in the same view. sqlGuard blocks writes, DDL, multi statement input and server file functions. Every query is logged and metered, credentials are encrypted, and keys can be revoked at any time.

Frequently asked questions about AI2SQL

  • How much does AI2SQL cost?
    Free plan includes 3 queries a day on a demo database. Paid tiers are Start at $9 per month, Pro at $19 per month and Team at $39 per month. Connecting your own database requires Pro or Team, and the Windows and macOS desktop app is a Pro feature.
  • Does AI2SQL offer a free trial?
    Every plan starts with a 7 day free trial. No charge is made on the first day, but a card is required to begin, and the trial can be cancelled in one click from the account before it ends.
  • Does AI2SQL have an API?
    Yes, AI2SQL offers an API.
  • How do you use AI2SQL?
    The walkthrough on this page covers 6 steps: 1. Create an account and open the builder 2. Connect a read-only database 3. Describe the result you want 4. Run it read-only and inspect the rows 5. Ask your database from your existing AI client 6. Run the connector next to a private database.
  • What platforms does AI2SQL support?
    AI2SQL is available on Web App, Windows and MacOS.
  • What does AI2SQL integrate with?
    Ten database connectors: PostgreSQL, MySQL, SQL Server, Oracle, Snowflake, BigQuery, Amazon Redshift, MongoDB, SQLite and Firebird. MCP works with Claude, ChatGPT in developer mode, Cursor and custom agents, and a REST API is also available.

Status

StatusActive
Views0
Outbound clicks0
Added10/22/2026

Platforms

Web AppWindowsMacOS

Pricing

FreemiumFree TrialSubscription