NOVA AI

Type a Question.
Get SQL That Works.

Schema-aware text-to-SQL that generates dialect-correct queries, validates against your real database, and executes on live data — all from a single question.

Schema-Aware SQL Live Execution 10+ Database Dialects Instant Charts

Try Nova AI Learn More

Text to SQL: How Nova AI Turns Questions Into Database Queries

Text to SQL converts plain English questions into executable SQL queries using AI. Instead of writing SELECT statements manually, you describe what data you need — “show me last month’s sales by region” — and the AI generates the correct query for your specific database. Schema-aware text-to-SQL tools go further by connecting to your actual database schema, validating every query against real table and column names, and executing against live data so you never have to copy-paste between tools.

Query Streams is a secure, real-time database integration platform that lets non-technical users query any database through natural language. Nova AI, the text-to-SQL engine inside Query Streams, generates dialect-correct SQL for 10 databases, validates every query through a multi-layer pipeline, and runs it on live data — all from one question. Learn more at QueryStreams.com and sign up to start querying.

How Text to SQL Works

Text-to-SQL is not one step — it is a pipeline. The quality of the output depends on how many validation layers exist between your question and the final result. Nova AI runs six stages for every query, from interpreting your intent to delivering results with optional charts and analysis.

Schema Analysis is the critical first step. Before generating any SQL, Nova pre-loads your actual table metadata through Schema Intelligence — table names, column names, data types, and relationships. This is what separates schema-aware text-to-SQL from generic AI that guesses at your database structure.

Multi-layer validation then checks every generated query: syntax checking ensures the SQL is valid for your specific dialect, table validation confirms every referenced table exists in your schema, column validation verifies every column name (with fuzzy-match correction for misspellings), and structural checks catch errors like invalid GROUP BY clauses or multi-statement queries. Only after passing all layers does Nova execute the query against your live database.

Why Schema-Aware Text to SQL Matters

Not all text-to-SQL tools are equal. Generic AI assistants like ChatGPT can generate SQL from a description, but they have no knowledge of your actual database. They guess at table names, invent column names that don’t exist, produce syntax that may not work on your specific database engine, and leave you to copy-paste the result into another tool to actually run it.

Schema-aware tools eliminate these problems entirely. Nova AI reads your database schema before generating anything, produces SQL in the exact dialect your database expects, validates every table and column reference against your real metadata, and executes directly on your live data. The result is not just a SQL string — it’s an answer, complete with data, charts, and natural language analysis.

For a deeper look at how natural language translates to SQL in practice, see Natural Language to SQL: How Nova AI Turns Questions Into Database Queries.

What Makes Nova AI Different

Nova AI is built specifically for database querying, not adapted from a general-purpose chatbot. Every component — from Schema Intelligence to the dialect engine — is designed for one purpose: turning questions into correct, executable SQL.

Supported Databases

Nova AI generates dialect-specific SQL for every database that Query Streams supports. Deploy multiple Agents across your infrastructure — on-premise servers, AWS EC2 instances, Azure VMs, Google Cloud, or any environment globally. Each agent auto-updates and makes its data sources available as a unified list in Nova’s interface.

Supported databases: Microsoft SQL Server (T-SQL), PostgreSQL, MySQL, MariaDB, SQLite, Microsoft Access, Oracle, Snowflake, BigQuery, and DuckDB. Nova also works with API Bridge connectors — data from Stripe, HubSpot, Shopify, Google Analytics, Google Search Console, and ShipStation is cached locally in DuckDB and queryable through natural language just like any other database. More integrations are coming soon.

Getting Started with Text to SQL

Going from zero to querying your database in natural language takes three steps. No SQL knowledge required.

Step 1: Install the Query Streams Agent. Sign up at my.querystreams.com and install the Agent on any machine that can reach your database. The Agent connects outbound only — no firewall changes, no inbound ports, no VPN configuration. It runs on Windows, macOS, and Linux.

Step 2: Open Nova AI. Navigate to Nova from your Query Streams dashboard and select a database and data source. Schema Intelligence automatically loads your tables, columns, data types, and relationships in the background. Nova is available on Business and Enterprise plans.

Step 3: Ask your question. Type a question in plain English: “Show me all orders from last quarter over $500, grouped by customer.” Nova generates dialect-correct SQL for your database, validates it through the multi-layer pipeline, runs it against live data, and presents the results. Ask follow-up questions, request charts, or refine the query through conversation.

Beyond Text to SQL: Save and Reuse

Text-to-SQL is the starting point, not the end. Once Nova generates a query you want to keep, you can save it and use it across the entire Query Streams platform:

Run in Microsoft Excel or Google Sheets. Saved queries appear in the Query Streams add-in alongside manually created queries. Select a query, adjust any available filters, and run it — live data streams directly into your spreadsheet on-demand. Query Streams supports parallel query execution, so you can launch multiple queries into different worksheet tabs simultaneously.

Share with teammates or external partners. Recipients run shared queries from their own Excel or Google Sheets. They can adjust filters independently — date ranges, regions, product categories — without ever seeing your SQL code or database credentials. You share the interactive capability to query data, not the data itself.

Add parameters for interactive filters. Convert static values in your query into dynamic filters. When you or your recipients run the query, they see dropdown menus, date pickers, and text inputs — not raw SQL. Query Streams detects both explicit variables and static values in your query that can become filters.

For more on how AI analytics extends beyond text-to-SQL, see AI Powered Analytics.

Who Benefits from Text to SQL

Frequently Asked Questions

Start Querying in Plain English

For more on Nova AI capabilities, see the feature page. For complete analytics workflows including charts, trend analysis, and data storytelling, see AI Powered Analytics. For database connection guides, see SQL Server to Excel, PostgreSQL to Excel, or MySQL to Google Sheets.