AI · Data Analysis · LLM · BigQuery

AI - Data Insight Generator

An AI-powered data analysis platform that connects to databases and data warehouses like BigQuery — letting users ask natural language questions and instantly receive insights, summaries, and auto-generated visualizations.

NL
Natural Language Queries
Plain English input
Auto
Chart Generation
Matplotlib & Seaborn
Live
BigQuery Connection
Real-time data
LLM
AI-Powered Insights
OpenAI powered

What We Built & Why

AI — Data Analysis is an intelligent analytics platform that bridges the gap between business users and raw data warehouses. Instead of writing SQL queries or waiting for BI reports, users simply ask questions in plain English — and the platform generates the query, executes it against BigQuery, and returns insights with auto-generated charts.

The platform eliminates the need for manual querying and report generation — reducing analysis turnaround from hours to seconds while making data exploration accessible to non-technical stakeholders.

Core problem solved: Business teams waste hours writing SQL, waiting for analysts, or navigating complex BI tools. This platform lets anyone ask "What is our monthly churn rate?" and receive an instant answer with a chart — powered by LLM + BigQuery + Python.

Natural Language

Ask complex questions in plain English — no SQL knowledge required.

Auto Visualizations

Charts and graphs auto-generated from query results — zero manual effort.

BigQuery Connected

Directly wired to data warehouses for live, real-time query execution.

AI Summaries

LLM generates concise, readable insight summaries from raw result data.

The AI Analysis Pipeline

Every query flows through an intelligent pipeline — from plain English input to SQL generation, live data execution, and auto-generated visual insights.

01
Natural Language Input

User types a question in plain English — e.g. "churn customers in December 2025" or "monthly total revenue".

02
LLM SQL Generation

The LLM converts the question into an optimised SQL query — schema-aware and tuned for BigQuery execution.

03
BigQuery Execution

SQL executed against the live BigQuery data warehouse. Results fetched in real time — no stale cached data.

04
AI Summary

LLM reads the raw result data and generates a concise, readable business summary with key takeaways.

05
Auto Visualization

Matplotlib / Seaborn selects and renders the best chart type — bar, line, pie, or scatter based on data shape.

06
Interactive Display

Results, summary, and chart rendered together in the UI. User can explore different queries in real time.

07
Follow-Up Queries

Users can refine or extend queries conversationally — "now break it down by region" builds on prior context.

08
Report Ready

Analysis complete — no manual querying, no BI tool configuration, no waiting for the data team.

See It in Action

Real screenshots from the deployed Data Insight Generator platform — from natural language query input to auto-generated SQL, query results, and interactive visualizations.

Data Insight Generator — Natural Language Query Interface
Natural Language Query Input

Users simply type a question in plain English — e.g. "show me the top 15 users by revenue". The platform connects directly to Google BigQuery as the live data source.

· No SQL knowledge required
· BigQuery selected as data source
· Dataset auto-detected from connection
LLM Generates SQL & Executes Query

The LLM converts the plain English query into optimised SQL with SELECT, ORDER BY, CAST, and LIMIT — executed against BigQuery in real time.

· Query executed successfully — green indicator
· AI explanation of what the query does
· Result table with live data returned instantly
Generated SQL query executed with results in BigQuery
Bar chart auto-generated with chart type selector
Auto-Generated Bar Chart Visualization

After query execution, the platform automatically renders the most suitable chart. A Bar Chart visualises the result data — with a live chart type switcher to change views instantly.

· Bar, Line, Scatter, Pie, Area available
· One-click chart type switching
· Interactive, zoomable, downloadable
Area Chart — Revenue Analysis

Switch to an Area Chart to visualise total_revenue across user_id — values up to 300K showing high-value customer concentration. Same data, different visual insight.

· Revenue distribution clearly visible
· Handles 45K+ user_id range seamlessly
· Download results button below chart
Area chart showing total revenue by user ID up to 300K

What Users Can Ask & Get

"Churn customers in December 2025"
· Returns count, customer IDs, revenue lost
· Churn date breakdown by day
· Trend chart vs prior periods
· Auto Bar Chart generated
"Churn rate by month"
· Calculates churn rate as a percentage
· Month-over-month with trend direction
· Highlights peak churn periods
· Auto Line Chart generated
"Monthly total revenue"
· Aggregates total and net revenue by month
· Highlights growth and decline periods
· Revenue trend comparison
· Auto Area Chart generated
LLM-Powered SQL Generation
· Schema-aware query construction
· Handles complex aggregations & joins
· Optimised for BigQuery performance
· Safe parameter injection — no SQL injection
BigQuery Integration
· Direct BigQuery API connection
· Supports multiple datasets & tables
· Real-time query execution — no cache
· Extensible to other data warehouses
Query Optimisation
· Generated SQL reviewed & optimised
· Partition pruning for BigQuery cost control
· Pagination for large result sets
· Query result caching for repeated runs

Who Uses It & For What

Business Leaders

Ask "What was our Q4 revenue?" or "Which region had the highest churn?" — instant board-ready insights.

Growth & Marketing

Track campaign performance, user acquisition funnels, and conversion rates — no analyst needed.

Finance Teams

Monthly revenue reconciliation, cost analysis, and forecast comparisons in seconds instead of hours.

Product & Ops

Analyse user behaviour, feature adoption, and operational KPIs interactively — with auto-generated dashboards.

What This Platform Delivers

Analysis time cut from hours to seconds — complex queries answered instantly without waiting for the data team.

Non-technical users can explore data independently — no SQL knowledge or BI tool training required.

Auto-generated charts eliminate manual reporting — every answer comes with a relevant, publication-ready visual.

Scales with your data — connected directly to BigQuery to handle millions of rows with optimised, cost-efficient queries.

Interactive reporting replaces static dashboards — users explore data dynamically and refine insights conversationally.

Technologies Used

Python
Core backend language
LLM / AI
NL to SQL & summaries
SQL
Query generation & execution
BigQuery
Data warehouse
Matplotlib
Chart rendering
Seaborn
Statistical visuals

What Was Built & Applied

Python
AI & Large Language Models (LLM)
SQL
BigQuery
Data Visualization (Matplotlib, Seaborn)
Data Analysis Automation
Query Optimization

Want AI-powered data analysis for your business?

From natural language queries to auto-generated charts — we build intelligent analytics platforms that make your data work for everyone.

Start Your Project →