Ask Your Database.
In Plain English.
Query any database with natural language instead of SQL. Schema-aware AI that understands your tables, writes dialect-correct queries, and returns results with charts — no code required.
Ask Nova, get SQL + charts
Meet Nova
One key per partner. No credentials shared.
Build an API
+more
Scheduled sync to 6+ platforms
Explore
+more
Query APIs with SQL, no code
Explore
+more
Claude, Cursor, ChatGPT & Grok talk to your data
Connect AIQuery Streams is a secure, real-time database integration platform that brings conversational BI to every database in your organization through Nova AI — a schema-aware engine that converts plain English questions into dialect-correct SQL queries with auto-generated charts. Learn more at QueryStreams.com and sign up for free to start querying your databases in natural language.
What Is Conversational BI?
Conversational BI is a category of analytics technology defined by Gartner as part of the broader augmented analytics trend. Instead of building dashboards, writing SQL queries, or navigating complex reporting tools, users ask questions in natural language — “What were our top-selling products last month?” or “How does this quarter’s revenue compare to last year?” — and the system interprets the question, queries the underlying database, and returns a direct answer with a visualization.
This represents a fundamental shift in how organizations interact with data. Traditional business intelligence requires specialized analysts, weeks of dashboard development, and ongoing maintenance every time a new business question arises. Conversational BI makes data access exploratory and immediate: anyone in the organization can ask any question at any time without waiting for a report to be built or a SQL query to be written.
Natural Language Input
Ask questions the way you naturally think about data, without learning SQL syntax or navigating BI tool interfaces.
Instant Answers
Get results in seconds instead of waiting hours or days for a report request to be fulfilled by an analyst or DBA.
Exploratory Analysis
Follow up with additional questions in the same conversation, refining your analysis in real time without starting over.
Accessible to Everyone
Business users, managers, and executives can query data without any SQL training, database knowledge, or technical background.
The promise of conversational BI is compelling: democratize data access across the organization, reduce the bottleneck on analysts and DBAs, and let decision-makers get answers when they need them — not when someone has time to write a query. But the technology behind it matters enormously. A conversational BI tool is only as good as the AI engine powering it, and most implementations fall short in critical ways.
Why Most “AI SQL” Tools Fail
Since the rise of large language models, dozens of tools have appeared claiming to convert natural language into SQL. The concept sounds straightforward: describe what you want, get a query back. In practice, most of these tools share a critical flaw — they do not know your database schema. A generic AI model can write syntactically correct SQL, but without knowledge of your actual table names, column types, relationships, and business conventions, the output is unreliable at best and dangerous at worst.
Consider what happens when you ask a generic AI to “show me revenue by product category.” The model might generate a query referencing a customers table when your table is named tbl_clients. It might assume a column called revenue exists when your schema stores that data as total_amount_usd across a junction of two tables. These are not edge cases — they are the normal outcome of asking a model to write SQL against a schema it has never seen.
Schema Hallucination
Generic AI invents table and column names that sound plausible but do not exist in your database, producing queries that fail on execution.
Dialect Blindness
SQL syntax differs significantly between PostgreSQL, SQL Server, MySQL, and Oracle. Generic tools often produce syntax invalid for your specific database engine.
No Context Retention
Most tools treat each question as independent, losing the conversational thread. Follow-up questions like “break that down by region” fail without prior context.
No Validation Layer
Queries go directly to execution without syntax checking, returning cryptic database errors to non-technical users instead of helpful guidance.
How Nova AI Powers Conversational BI in Query Streams
Nova AI is the conversational BI engine built into Query Streams. Unlike standalone AI SQL generators, Nova connects directly to your database schema through the Query Streams Agent and generates queries against your actual table structures, column names, and data types. This is not a generic model guessing at your schema — it reads the real thing before writing a single line of SQL.
- Schema Intelligence: Nova inspects your live database schema — tables, columns, data types, primary keys, and foreign key relationships — before generating any query. Every reference in the output SQL maps to a real object in your database.
- Dialect-correct output: Nova generates SQL that matches the syntax of your specific database engine, whether that is PostgreSQL, SQL Server, MySQL, Oracle, Snowflake, BigQuery, SQLite, MariaDB, or DuckDB.
- Auto-generated charts: Every query result can be accompanied by an automatically generated visualization. Nova supports line charts, bar charts, pie charts, scatter plots, area charts, and horizontal bar charts.
- Conversation context: Nova maintains context across a conversation, so follow-up questions like “break that down by region” or “now show only Q4” work naturally without restating the entire original question.
- Credit-based usage: Nova operates on a credit system, giving organizations predictable cost control over AI-powered analytics. Credits can be purchased through the Query Streams portal, and usage is tracked per user.
- Query validation: Before execution, Nova validates generated SQL for syntax errors and safety, catching problems before they reach your database and returning clear guidance instead of cryptic error messages.
When combined with Query Streams’ saved queries, the Microsoft Excel add-in, and the Google Sheets add-on, Nova completes the picture. Pre-built saved queries handle recurring reports, while Nova handles ad-hoc exploration. Together, they form a comprehensive sql query builder online that serves both technical users who prefer to write their own SQL and business users who would rather type a question in plain English.
Schema Intelligence: The Foundation of Accurate Queries
Schema Intelligence is the feature that separates Nova from generic text-to-SQL tools. When you connect a data source through the Query Streams Agent, Nova reads the schema metadata — every table, every column, every data type, every relationship — directly from the database engine’s system catalogs. This happens automatically, without any manual configuration.
This is fundamentally different from tools that ask you to paste your schema into a chat window or upload a CSV of table definitions. Those approaches are static: the moment your schema changes — a new column added, a table renamed, an index created — the AI’s understanding is out of date. Nova reads the live schema on demand, so it always reflects the current state of your database. If your team adds a customer_lifetime_value column to the accounts table on Monday, Nova knows about it on Tuesday without any manual intervention.
Schema awareness also means Nova can generate JOINs correctly. If your customers table has a foreign key to orders, and orders links to products, Nova knows those relationships exist and can write multi-table queries without you specifying how the tables connect. For organizations with complex data models spanning dozens or hundreds of tables, this eliminates a significant source of error that plagues generic AI SQL tools and even experienced analysts writing queries by hand.
Dialect-Aware SQL Generation
One of the most overlooked challenges in natural-language-to-SQL tools is dialect variation. SQL is not a single language — it is a family of dialects, and the differences between them are significant enough to break queries entirely. A query that runs flawlessly on PostgreSQL will produce a syntax error on SQL Server. A function that exists in MySQL has a completely different name and behavior in Oracle. Any conversational BI tool that ignores these differences will fail the moment it encounters a database engine it was not specifically trained on.
| Operation | PostgreSQL | SQL Server | MySQL | Oracle |
|---|---|---|---|---|
| Limit results | LIMIT 10 |
TOP 10 |
LIMIT 10 |
FETCH FIRST 10 ROWS ONLY |
| Current timestamp | NOW() |
GETDATE() |
NOW() |
SYSDATE |
| String concatenation | || |
+ |
CONCAT() |
|| |
| Identifier quoting | "column" |
[column] |
`column` |
"COLUMN" |
| Boolean type | BOOLEAN |
BIT |
TINYINT(1) |
NUMBER(1) |
| Date extraction | EXTRACT(MONTH FROM d) |
MONTH(d) |
MONTH(d) |
EXTRACT(MONTH FROM d) |
Nova handles all of these variations automatically. When you ask “show me the top 10 customers by revenue,” Nova knows whether to generate TOP 10 (SQL Server), LIMIT 10 (PostgreSQL, MySQL), or FETCH FIRST 10 ROWS ONLY (Oracle) based on the database engine your data source uses. This dialect awareness extends to hundreds of syntax differences across string functions, date handling, aggregation, window functions, type casting, and more.
For organizations that run multiple database platforms — a common scenario in enterprises with legacy SQL Server systems alongside newer PostgreSQL deployments, or companies using both MySQL for applications and Snowflake for analytics — this means a single conversational BI interface that works consistently across every database. Users do not need to know which engine stores the data they are querying, and they certainly do not need to know the syntax differences between them.
From Question to Chart: The Full Nova Workflow
The conversational BI workflow in Nova is designed to be frictionless. A user types a question, and within seconds they have both the data and a visualization. Behind the scenes, Nova performs a series of steps to ensure accuracy — but from the user’s perspective, it feels like a conversation with someone who already understands the database.
Ask Your Question
Type a natural language question in the Nova chat interface, such as “What were our top-selling products last quarter?”
Schema Analysis
Nova reads the connected database schema, identifying relevant tables, columns, data types, and foreign key relationships.
SQL Generation
Dialect-correct SQL is written automatically, matching the syntax and functions of your specific database engine.
Validation and Execution
The generated query is validated for syntax and safety, then executed through the Query Streams Agent against your live database.
Results and Visualization
Data results return alongside an auto-generated chart. Choose from line, bar, pie, scatter, area, or horizontal bar.
Nova’s chart generation selects the most appropriate visualization based on the data structure returned. Time-series data gets a line chart. Category comparisons get a bar chart. Distribution analysis gets a scatter plot. You can switch between the six available chart types — line, bar, pie, scatter, area, and horizontal bar — after the query runs if you prefer a different view of the same data.
Because Nova maintains conversation context, you can immediately follow up: “Now show me that as a pie chart,” “Filter to just the North America region,” or “What does that look like month over month?” Each follow-up builds on the previous context, making the experience genuinely conversational rather than a series of isolated queries. The full conversation history is preserved, so you can revisit and continue any analysis from where you left off.
Conversational BI vs Traditional BI
Traditional BI and conversational BI serve different but complementary purposes. Understanding where each approach excels helps organizations build an analytics strategy that covers both recurring reporting needs and ad-hoc exploration. The comparison below highlights the fundamental differences in approach, speed, and accessibility.
-
Dashboard dependency
Every new insight requires a pre-built dashboard or report, meaning weeks of lead time before a new metric is available.
-
Specialist bottleneck
BI analysts and data engineers control the pipeline, creating a queue for every request from business stakeholders.
-
Static perspectives
Dashboards answer only the questions they were designed for. If the business question changes, someone must rebuild the dashboard.
-
High licensing costs
Enterprise BI platforms typically charge per-seat licenses that can run tens of thousands of dollars annually for mid-size teams.
-
Ask anything, anytime
Any question gets an immediate answer with a visualization. No dashboard to build, no report to request, no queue to wait in.
-
Zero specialist dependency
Business users query data directly through natural language. The AI handles SQL generation, validation, and chart creation.
-
Exploratory by design
Every conversation is a fresh exploration. Follow-up questions refine the analysis in real time, enabling discovery of unexpected insights.
-
Predictable credit pricing
Nova’s credit system means you pay for what you use, without per-seat BI platform licenses or annual enterprise contracts.
This does not mean conversational BI replaces traditional BI entirely. Pre-built dashboards remain valuable for KPIs that need daily monitoring, executive scorecards, and operational metrics that should refresh automatically on a schedule. Conversational BI excels at the ad-hoc, exploratory questions that fall outside your existing dashboards — the questions you cannot predict in advance. The strongest analytics strategy combines both: dashboards for the known questions, and conversational BI with Nova for everything else.
Who Conversational BI Is For (and When You Still Need SQL)
Conversational BI is not one-size-fits-all. Different roles within an organization benefit in different ways, and there are legitimate scenarios where writing SQL directly remains the better approach. Understanding both sides helps you deploy Nova where it delivers the most value.
Business Analysts
Analysts who know their data but want faster ad-hoc queries. Nova accelerates the “quick question” workflow that currently requires opening a SQL client and writing code from scratch.
Non-Technical Users
Sales managers, marketing leads, and finance teams who need data answers but have no SQL knowledge. Nova translates their plain English questions into database queries automatically.
Executives and Decision Makers
Leaders who need quick answers to strategic questions — revenue trends, customer metrics, operational KPIs — without scheduling a report from the analytics team.
Small Teams Without BI Staff
Organizations that cannot afford a dedicated BI team but still need data-driven insights from their SQL Server, PostgreSQL, MySQL, Oracle, or other databases.
The practical sweet spot for conversational BI is the daily volume of “can you pull this data?” requests that currently land on your analysts’ or DBAs’ desks. Revenue reports, customer lists, product performance summaries, vendor spend breakdowns, headcount by department — these are the questions that take an expert five minutes to answer but create a two-day ticket queue when thirty people ask them in the same week. Nova handles these instantly, freeing your technical team for the complex work that actually requires their expertise.
Frequently Asked Questions
What exactly is conversational BI? +
How is Nova AI different from using ChatGPT to write SQL? +
Which databases does Nova support for conversational BI? +
Do I need to manually describe my database schema to Nova? +
What types of charts can Nova generate automatically? +
How does the Nova credit system work? +
Can Nova handle complex queries with JOINs and subqueries? +
Is my database data sent to external AI servers? +
Can I save and reuse queries that Nova generates? +
Get Started
Start Querying in Plain English
Try conversational BI with Nova AI. Connect your database, ask a question, and get answers with charts — no SQL required. Free tier available with credits included.
Related guides: AI Data Analytics with Nova | Text to SQL | Databases in Microsoft Excel | Databases in Google Sheets | Download the Query Streams Agent
Category: Articles
Tags: conversational BI, natural language SQL, AI data analytics, text to SQL, Nova AI, AI query builder, no-code analytics, plain English database, schema-aware AI, query streams
Meta Description: Query any database in plain English with conversational BI. Schema-aware AI that writes SQL for you.
English 
Spanish 
German 
French 
Portuguese 
Japanese 
Korean 
Chinese 
Italian 
Dutch 
Indonesian 
Polish