Forem: Sanjay_Balaji

Building an AI-Powered SQL Chatbot With LangChain, Mistral & Streamlit

Sanjay_Balaji — Sun, 30 Nov 2025 15:14:23 +0000

(Query your PostgreSQL database using natural language)

Managing a growing database can get messy — remembering table names, joins, and Sequelize-generated column structures isn’t fun.
So I built an AI-powered SQL Chatbot that turns natural language questions into safe, optimized SQL queries, and streams results instantly.

This project uses:

LangChain + LangGraph
Mistral AI (mistral-small-latest)
PostgreSQL + Sequelize ORM
Streamlit UI

What This Chatbot Does

Simply type something like:
“Show me the last 5 orders with customer names.”
The chatbot will:
Parse the question

Understand the relevant tables
Auto-generate a safe, dialect-correct SQL query
Avoid soft-deleted rows
Perform the right JOINs using Sequelize relationships
Stream the result back to you in seconds
No manual querying. No database digging.

What This AI Chatbot Can Do

✔ Understand natural language questions
✔ Generate accurate and safe SQL
✔ Join across multiple related tables
✔ Auto-fix incorrect SQL
✔ Enforce business rules (soft-delete filters, safe SELECT-only mode)
✔ Display results interactively in Streamlit

Architecture at a Glance

User asks a question – “What prepaid services does customer John have?”
LangChain SQL Toolkit analyzes table schemas
Mistral LLM writes an optimized SQL query
PostgreSQL executes it
Streamlit shows the answer cleanly

Simple. Powerful. Zero SQL typing

Tech Stack

Component	Purpose
Mistral AI (`mistral-small-latest`)	Generates SQL using natural language
LangChain SQLToolkit	Provides DB inspection & SQL execution
LangGraph ReAct Agent	Tool-based reasoning loop
PostgreSQL	Database
Streamlit	Lightweight chat UI

Core Code Snippet

Here’s the essential logic behind the agent:

llm = ChatMistralAI(model="mistral-small-latest", api_key=api_key)

toolkit = SQLDatabaseToolkit(db=db, llm=llm)
tools = toolkit.get_tools()

agent = create_react_agent(llm, tools, prompt=system_prompt)

for step in agent.stream({"messages": [{"role": "user", "content": question}]}, stream_mode="values"):
    last_msg = step["messages"][-1]
    if isinstance(last_msg, AIMessage):
        st.markdown(last_msg.content)

The SQL agent:

inspects schemas
constructs joins
limits output
blocks dangerous commands
retries if SQL errors Pretty smart for 20 lines of code

Why This Is Useful
This chatbot is perfect for:

internal dashboards
support teams
BI analysts
non-technical managers
quick debugging
ad-hoc data exploration

It turns your database into a chat interface, which team members can use instantly — no SQL skills required.

Try It in Your Project
If you’re building tools that query relational data, this setup is incredibly powerful.
Plug in your DB → set your rules → get a natural-language assistant instantly.

Automating My Daily Workflow with n8n, Python & Gemini AI

Sanjay_Balaji — Sun, 30 Nov 2025 14:49:12 +0000

How I built smart automations that save hours every week

Automation doesn't need huge infrastructure — sometimes all you need is n8n + Python + a good LLM.
Here’s how I built three lightweight but powerful automations that improved my content flow and daily routine.

1. Auto-Generate LinkedIn Tech Posts (gNews → Python → Gemini → LinkedIn API)

I wanted fresh, relevant tech content posted automatically on LinkedIn — without doing it manually every morning.

Workflow Includes:

Fetching real-time tech news via gNews API
Cleaning + formatting content in Python
Generating a polished LinkedIn-ready caption using Gemini AI
Publishing automatically through LinkedIn API

Result:
✔ Zero manual posting
✔ Consistent daily visibility
✔ Cleaner + AI-improved content

2. Daily Weather + Motivation Telegram Bot

A simple automation that sends a morning weather report and a Gemini-generated motivational message.

Workflow Includes:

OpenWeatherMap API → real-time weather
Gemini AI → custom motivation text
Telegram Bot API → scheduled daily message

Result:
✔ Daily personalized message
✔ A small workflow that boosts productivity
✔ Fully autonomous once deployed

Why These Automations Matter

They remove repetitive tasks
They help maintain consistent personal branding
They show how flexible n8n + Python + Gemini AI can be
They are small, modular, and easy to scale

Final Thoughts

If you’re looking to automate small but impactful tasks, start with tiny workflows — news fetching, posting, messaging, alerts — and build upward.
n8n + Python + Gemini AI turned out to be the perfect trio for me.

If you want the full workflow JSON or code snippets, drop a comment — happy to share!

LogInSight: A Lightweight CloudWatch Log Analytics Tool for Faster Debugging and Real-Time Insights

Sanjay_Balaji — Sat, 29 Nov 2025 15:54:10 +0000

Modern applications generate thousands of logs every day, but teams still struggle to find issues quickly. Traditional log monitoring often involves searching manually through CloudWatch, jumping between services, and spending valuable time correlating events.
LogInSight solves this problem by centralizing log ingestion, organizing logs in a lightweight backend, and offering instant analytics through a clean, interactive dashboard.

This article explains how LogInSight improves debugging efficiency and how the system works from log ingestion to real-time insights

How LogInSight Improves the Business

Centralized Log Visibility
Instead of toggling between multiple CloudWatch log groups, LogInSight brings everything into a unified dashboard.

Faster Debugging and Issue Detection
Engineers can instantly view errors, warnings, request failures, and event spikes without manually filtering logs.

Better Operational Insights
Log analytics, error frequency charts, timestamps, and stream summaries help teams understand system behavior at a glance.

Lightweight and Cost-Efficient
LogInSight uses Python, FastAPI, SQLite, and Streamlit—keeping hosting and maintenance extremely simple.

Real-Time Awareness
AWS log polling enables near real-time updates, helping teams react before issues escalate.

Log Ingestion from AWS CloudWatch
LogInSight connects to CloudWatch using AWS SDKs and pulls logs from configured log groups.The ingestion pipeline:

Retrieves logs in batches
Converts raw streams into structured JSON
Stores processed entries in a local SQLite database
Categorizes logs by service, level, and timestamp

This creates a consolidated log source that is easier to search and analyze.

Future-Ready: From Static Logs to Live Log Streaming
The current version of LogInSight is built to process static CloudWatch log batches, but the architecture is designed to scale into a real-time log streaming system with minimal changes.
By integrating:

AWS Lambda to trigger whenever new CloudWatch logs arrive
Amazon Kinesis to stream logs continuously
A Python log consumer to process live data
Streamlit to update the dashboard automatically

LogInSight can evolve into a fully real-time observability platform, delivering instant insights during outages, deployments, or high-traffic periods.

Structured Storage With SQLite

Logs are stored in a clean and lightweight schema:
Log message
Timestamp
Log level (INFO, WARN, ERROR)
Service / log group

Additional metadata

SQLite keeps the system simple, fast, and portable—ideal for teams that want quick deployment without heavy infrastructure.

Backend API Powered by FastAPI
FastAPI serves as the analytics engine behind LogInSight.The API provides endpoints for:

Fetching logs based on filter criteria
Searching logs by keywords
Retrieving error summaries
Fetching time-series data for charts
Providing service-wise log breakdowns

This backend ensures low-latency responses even with large log volumes.

Real-Time Analytics Dashboard (Streamlit)
The interactive dashboard gives teams a clear, actionable view of system health. It includes:

Live log stream viewer
Error and warning frequency charts
Service-level log summaries
Time-based filtering
Search and pattern matching
Highlighting for critical log events

This enables teams to pinpoint issues instantly and understand trends with minimal effort.

For the complete implementation, check my GitHub repository:

Why LogInSight Matters
LogInSight is designed for teams who need powerful log visibility without complex tooling. It provides:

Unified view of distributed application logs
Faster troubleshooting during outages
Clean dashboards for engineering and DevOps teams
Predictable insights into system behavior
Low-resource log analytics without enterprise-level overhead

It transforms raw logs into meaningful operational intelligence—improving uptime, reliability, and debugging speed.

Smart Invoice Analyzer — How I Automated Invoice Processing & Predicted Sales Using Machine Learning

Sanjay_Balaji — Sat, 29 Nov 2025 13:21:09 +0000

Modern businesses process thousands of invoices, but many still rely on manual entry. This slows down operations, introduces errors, and hides important sales trends. The Smart Invoice Analyzer solves these problems by automatically extracting invoice data, organizing it into meaningful structures, forecasting next month’s demand, and presenting insights through a clear, interactive dashboard.

This article shows how the system improves business efficiency and walks through the workflow from invoice upload to final prediction.

How Smart Invoice Analyzer Improves the Business

Faster and More Accurate Invoice Processing
Manual entry is time-consuming and prone to errors. Automated extraction eliminates repetitive work and ensures greater accuracy across all invoice data.

Clear Visibility Into Monthly Sales Trends
The system provides instant visibility into revenue, units sold, product categories, and monthly fluctuations. Businesses no longer need to depend on scattered spreadsheets to understand performance.

Smarter Inventory and Procurement Planning
A built-in machine learning model predicts next month’s top-selling product. This helps businesses maintain optimal stock levels, avoid shortages, and reduce excess inventory.

Better Decision Making Across Departments
Finance, operations, and retail teams work from the same structured dataset and dashboard, enabling consistent and data-driven decision-making.

Uploading Invoices
Users upload a ZIP file containing all invoices for the month.
Once uploaded, the system automatically:

Extracts the files
Reads each invoice
Identifies key fields
Processes everything without manual work

This creates a predictable, repeatable workflow every month.

Automated Data Extraction Using Python
The extraction pipeline automatically:

Reads each invoice PDF
Detects and interprets important details
Cleans and standardizes the extracted information
Converts unstructured data into a structured format
Ensures consistency across invoices from different templates
Prepares the data for storage, analysis, and forecasting

This turns messy PDFs into clean, structured data.

Predicting Next-Month’s Demand (Machine Learning)
The forecasting module uses an XGBoost regression model. It:

Groups sales by product and month
Creates time-based features (month, year, time index)
Trains a model for each product
Predicts next-month quantity
Identifies the highest-selling upcoming product

This makes procurement and inventory planning proactive instead of reactive.

For more details, check my GitHub repository:
https://github.com/sanjay-dot/smart-Invoice-Analyzer

Dashboard Insights for Executives
Once the invoices are processed, the dashboard provides:

A consolidated view of sales and revenue
Clear month-to-month performance trends
Insight into how each category contributes to overall sales
Identification of top and slow-moving products
A breakdown of revenue concentration across product groups
Actionable metrics that support faster and more confident decision-making

These insights help teams make decisions quickly and confidently.

Why This System Matters

The Smart Invoice Analyzer is designed for small to mid-sized businesses that manage diverse inventories and significant invoice volume.
By combining automation, machine learning, and visual reporting, it delivers:

Reduced manual workload
Structured financial data
Accurate demand forecasting
Timely monthly reporting
Better operational and financial visibility

It transforms ordinary invoices into valuable business intelligence, enabling teams to make confident and data-driven decisions.