InsightLoop
About InsightLoop

Secure, Browser-First Business Intelligence

InsightLoop was designed to solve a fundamental challenge in modern business intelligence: analyzing sensitive data quickly and visually without compromising data privacy.

Why InsightLoop?

Standard BI tools require uploading spreadsheets to external cloud databases for ingestion, rendering, and reporting. For many organizations, analysts, and independent researchers, transferring sensitive records is either restricted by compliance or introduces unnecessary exposure.

InsightLoop shifts the entire analytical pipeline directly to your local sandbox. By utilizing next-generation browser-based technologies, your raw spreadsheet data remains fully local in your browser memory. We sync only the layout metadata, manual column types, and chat session outlines to the cloud, ensuring raw dataset rows never leave your computer.

About the Builder

InsightLoop was designed, engineered, and optimized single-handedly by Ahmad Khan, an aspiring Computer Science student and software engineer deeply focused on secure data processing, local-first web applications, and generative AI compilation pipelines.

Developed as a high-fidelity portfolio and engineering project, InsightLoop serves to demonstrate key architecture solutions:

  • Client-Side SQL Processing: Integrating DuckDB-WASM lazily to execute analytics queries at close to native speeds directly in the browser sandboxed environment.
  • AI Text-to-SQL Self-Healing Pipeline: Engineering robust query interfaces using the official Google Gemini 2.5 Flash SDK, complete with a client-side self-correction and retry mechanism when SQL errors are compiled.
  • Serverless Statistical Computations: Linking serverless Python endpoints to execute secure Outlier Detection, Regression Forecasting, and Correlation matrices without keeping or saving user records.

Privacy & Data Handling Commitment

Trustworthiness is our core foundational architecture. We guarantee the following data privacy policies, validated directly by our open-source codebase:

In-Memory Execution Only

Your uploaded CSV and Excel data is processed strictly inside your temporary browser tab memory. It is never uploaded, persisted, or written to any remote cloud disk.

Metadata-Only AI Inferences

When querying the AI Co-Pilot, we only transmit column names and data types (schema) to the serverless Gemini endpoint. Raw spreadsheet rows are never sent to the AI model.

Layout-Only Persistence

When saving a dashboard layout, only your customized widget configurations, SQL statements, and column display schemas are saved to Supabase. Raw dataset cells are never synchronized.

No Subscriptions or Tracking

We do not track you, sell advertising, or require any registration. Access is anonymous and managed using a highly secure client-side cookie rotation.

How It Works

01. Secure Parsing

Local Spreadsheet Ingestion

Drag & drop CSV or Excel files. Parsed instantly in-browser with zero server-side storage.

02. Instant Layouts

Automatic Dashboards

Interactive visualizations, timelines, and metrics build themselves instantly based on your data columns.

03. AI Co-Pilot

Conversational SQL

Ask business questions in plain English. The AI generates safe SQL queries executed locally in your browser database.

04. Export & Sync

PDF Reports & Reactivations

Compile beautiful PDF reports locally, and save secure shareable configurations to revisit later.

Built With

Next.js 15 & React 18Powering a responsive, server-side rendered application with smooth client transitions and modern app routing.
TypeScriptEnsuring strong type safety and robust development consistency across all components.
Google Gemini 2.5 APITranslating natural language queries into safe, highly optimized analytical SQL on the fly.
SupabasePersisting non-sensitive layouts, configurations, and conversation flows safely with zero data row exposure.
DuckDB-WASMThe high-performance, in-browser analytical database engine powering local-only SQL executions.
VercelProviding low-latency hosting and hosting serverless Python endpoints for heavy statistical computations.

Designed, engineered, and built with high professional authority by Ahmad Khan. Built with a deep commitment to data privacy, performance, and responsive product execution.