AutoTaranga, Bhaberchar, Gazaria (2026)

15/01/2026

Most businesses receive messages. Smart businesses respond intelligently, instantly, and at scale.
That’s exactly what this end-to-end Messenger Automation is built for.

Today, customers expect:
-Instant replies
-Accurate answers
-Support for both text and image queries
-Consistent experience — 24/7

This AI-powered messenger agent:
-Understands user text questions
-Analyzes user-submitted images
-Retrieves accurate answers from a knowledge base
-Responds automatically — without human intervention.
-Reduces support cost and response time drastically.

In short: better customer experience + scalable automation.

How I built this automation (high-level):
Using n8n as the orchestration layer, I designed a modular workflow:
-Webhook trigger receives Messenger input (text or image)
-Smart routing logic detects message type
-Image Analyzer Agent extracts only verified visual information
-Summarizer Agent converts image insights into concise summaries
-RAG Agent queries a Pinecone vector database for factual answers
-LLM layer responds strictly based on retrieved knowledge
-HTTP response sends the final reply back to Messenger
Every step is optimized for accuracy, reliability, and zero hallucination.

This is perfect for:
-Businesses with high Messenger traffic
-Customer support teams
-E-commerce & service platforms
-Founders who want AI without complexity

It’s a production-ready AI support agent designed to think, retrieve, and respond like a well-trained human, but faster.
If you’re exploring AI-powered messaging, RAG systems, or n8n automation, or want to automate your business, let’s connect.

14/01/2026

I’ve just completed a production-ready RAG (Retrieval-Augmented Generation) AI Agent workflow designed for businesses that need accurate, scalable, and automated knowledge-based responses without relying on static chatbots or manual support teams.
This system is built to solve a very real problem:
❌ LLMs hallucinate when they don’t have access to your data
❌ Support teams don’t scale
❌ Knowledge lives in docs, not in workflows
This RAG architecture fixes all three.

From a business and engineering standpoint, this system delivers:
Context-aware AI responses grounded in your own documents
Zero manual intervention after setup
Email-based AI support that works 24/7
Auditable, logged interactions for analytics and improvement
In short: your data becomes an API-driven AI brain.

System Architecture:
1️⃣ Knowledge Ingestion Pipeline
Source: Google Drive (documents as single source of truth)
Text chunking + embeddings generation
Vectors stored in a vector database (Pinecone / Qdrant / Weaviate)
One-time ingestion, continuously reusable
2️⃣ User Interface Layer
Built with Lovable AI (form-based frontend)
Input fields:
Name
Email
Phone
Query
Form submission triggers an n8n webhook
This keeps the UX simple while the backend remains powerful.

3️⃣ RAG Orchestration (n8n)
Webhook receives user query
Query converted into embeddings
Vector DB similarity search retrieves relevant context
LLM (gpt-4o-mini / Gemini) generates a grounded response using retrieved data
This is true RAG, not prompt stuffing.

4️⃣ Output & Observability
AI-generated answer sent directly to the user’s email
Admin notification for monitoring
Full query + response logged into Google Sheets / database for:
QA
Analytics
Future model improvements

AI agents are not about “using an LLM.”
They are about system design:
-Data pipelines
-Retrieval accuracy
-Deterministic workflows
-Observability and logging
When these are done right, AI becomes infrastructure, not a toy

This architecture is ideal for:
-SaaS companies
-Knowledge-heavy businesses
-Support automation
-Internal AI assistants
-Client-facing AI products
-If your business relies on documents, FAQs, policies, or internal knowledge this is how you scale it with AI safely.

If you’re interested in:
-Custom RAG systems
-AI agents powered by your data
-Production-grade AI automation
Let’s connect.

09/01/2026

Why every business needs a Market Research Agent (and why I built one)
Market research is the backbone of any successful business.
But let’s be honest—manual research is slow, repetitive, and exhausting.
Imagine:
⏺️Searching competitors one by one
⏺️Reading hundreds of Google reviews
⏺️Organizing messy data
⏺️Writing insights manually
This easily takes hours or even days.

So I built an AI-powered Market Research Agent using n8n that automates the entire process—from data collection to insight delivery end to end.
How this Market Research Agent helps businesses
⏺️Saves massive time ⏱️
⏺️Gives structured, actionable insights
⏺️Helps founders validate ideas faster
⏺️Supports better decision-making using real customer reviews
⏺️Scales research without hiring extra people.

How I designed the workflow (step-by-step)
1️⃣ Form Submission Trigger
The workflow starts when a user submits a form with their business or market input (location, niche, keywords, etc.).
2️⃣ User Input Configuration
The inputs are cleaned and structured so the system understands exactly what kind of market data is needed.
3️⃣ Google Maps Search
Using those inputs, the agent searches Google Maps to find relevant businesses and competitors.
4️⃣ Custom Code – Data Rearrangement
Raw data is messy. I use custom code nodes to restructure and normalize the results for further processing.
5️⃣ Loop Over Items
Each business is processed one by one to ensure no data is missed.
6️⃣ Review Extraction & Processing
The agent fetches customer reviews, analyzes the content, and applies custom logic to clean and prepare it.
7️⃣ Loop Ex*****on & Control
The loop runs efficiently until all businesses and reviews are processed.
8️⃣ Data Aggregation
All insights are combined into a single, structured dataset.
9️⃣ Final Custom Code Layer
I refine the aggregated data so it’s ready for AI analysis.
🔟 AI Agent (Gemini Model)
The AI analyzes patterns, sentiment, pain points, and opportunities—and generates human-like insights.
1️⃣1️⃣ Automated Gmail Response
Finally, the user receives a clear, actionable market research report directly in their email—no manual work needed.

This project reflects how I approach automation:
not just connecting tools, but designing intelligent systems that solve real business problems.
If you’re a founder, marketer, or business owner and want custom AI automation or agents for your business, feel free to DM me.
Let’s build systems that work while you sleep.

08/01/2026

Your website is the first impression of your business—make it count

Many people think a website is just about design.
But a great WordPress website is about speed, clarity, trust, and conversion.

I’m a WordPress Web Developer with a strong background in AI automation, and I help businesses build websites that are:

✅ Clean & modern
✅ Mobile-friendly
✅ Fast & SEO-optimized
✅ Easy to manage
✅ Built to generate leads, not confusion

From business websites to personal brands, portfolios, and landing pages, I focus on creating websites that actually work.

No unnecessary plugins.
No messy layouts.
No copy-paste themes.

Just a professional WordPress website tailored to your goals—with smart automation if needed (forms, emails, workflows).

If you’re planning to:
• Launch a new business
• Upgrade your old website
• Build a strong online presence

📩 Inbox me to discuss your WordPress project.
Let’s turn your website into a real business asset.

06/01/2026

I just built a full Customer Support RAG AI Agent from documents to real-time answers.
Not a concept. Not a prototype. This workflow is running accurately.

Here’s what’s happening behind the scenes 👇
I connected Google Drive as the main knowledge hub. Whenever a document is added or updated, the system automatically:
➡️Downloads the file
➡️Processes it with a default data loader
➡️Creates embeddings using OpenAI
➡️Stores everything inside a Pinecone vector database
This means the AI always stays updated with the latest business knowledge.
Then comes the customer support side 💬
In a separate workflow:
➡️The system receives a chat message
➡️Calls the AI agent
➡️Uses an OpenAI model with simple memory
➡️Retrieves the most relevant context from Pinecone
➡️Generates accurate, grounded responses using RAG (Retrieval-Augmented Generation)
So instead of hallucinating, the AI answers based on real company documents, policies, FAQs, guides, anything.
Why this matters for businesses 👇
✅ 24/7 customer support
✅ Faster response time
✅ Consistent answers
✅ Less pressure on human agents
I’m building and sharing these systems publicly because real automation creates real value, not just hype.

👉 If you’re a business owner or founder and want to automate your customer support or internal processes using AI, follow me and send me a DM.
Let’s see how AI can actually work for your business.

06/01/2026

Most people think AI = “trained once, then answers everything.”
That’s not true anymore.

Meet RAG - Retrieval-Augmented Generation.

🧠 What is RAG?
RAG is a technique where AI doesn’t rely only on what it learned during training.
Instead, it:
1️⃣ Retrieves fresh, relevant data from your own documents, databases, or knowledge base
2️⃣ Then generates answers using that data + its language skills

Think of it like this:
❌ Old AI = memory only
✅ RAG AI = memory + real-time research

💡 Why RAG is powerful

Reduces hallucinations

Keeps answers up-to-date

Lets you build private, custom AI (company data, PDFs, Notion, Airtable, etc.)

Much cheaper than retraining a model

⚙️ Where RAG is used

Customer support bots

Internal company AI assistants

Learning roadmap generators

Knowledge-based chatbots

📌 Simple idea, massive impact:
RAG turns AI from a “smart parrot” into a real problem-solver.

If you’re building AI products, RAG is not optional anymore — it’s the foundation.

Follow for more practical AI & automation insights.

04/01/2026

Just shipped something I’m genuinely excited about.
I’ve built a chatting AI agent that works like ChatGPT / Gemini but fully customized using n8n + an LLM + a custom frontend.
Here’s the full process in simple terms 👇

I started with n8n and set up a Webhook as the entry point. Every user message first hits this webhook.
Then I connected an LLM model, trained it with my own logic and instructions so it behaves exactly the way I want, not generic, not robotic.
After that, I designed a ChatGPT-style interface using Lovable AI app, focusing on clean UX and real-time interaction.
Whenever a user sends a message from the interface, it triggers the webhook → the AI processes it → and Respond to Webhook sends the reply back instantly.
No manual work. No copy-paste. Fully automated.
This setup can be used for support bots, internal tools, learning assistants, or business automation.

You can try the agent here:
👉 https://lnkd.in/gbHGxCB9

This project helped me understand how powerful AI automation + no-code tools can be when combined properly.
More experiments coming soon 👀

04/01/2026

Today I completed a project I’ve been thinking about for a while, a fully automated social media post generator built with n8n.

The workflow is simple, but the impact is huge:
A user submits a form → n8n triggers the workflow → information is extracted and structured → post content is prepared → and the post is automatically published to Facebook and LinkedIn. You may also published on instagram/TikTOk or any other social platforms.
No manual writing every time.

No copy-pasting between platforms.
Just one submission and the system handles the rest.

This project helped me understand how powerful n8n can be when it comes to building real-world automation, not just basic workflows. Connecting form submissions, data extraction, and social media publishing into one clean flow was a great learning experience.
What excites me most is how this kind of system can save time for creators, founders, and teams who struggle to stay consistent on social media.

It is clear that Automation doesn’t replace creativity, it removes friction so creativity can scale.
Still improving the workflow and adding smarter logic step by step.

02/01/2026

Recently, I built an AI agent in n8n that can actually think, search, filter, and act, not just trigger workflows.

Here’s what I did, step by step 👇
I started by creating an n8n agent that executes workflows dynamically.
Then I trained the agent using a customer datastore inside n8n, so it understands what kind of data it should work with and how to respond intelligently.
After that, I connected Airtable as the main data source.
Before pushing anything into Airtable, I:
Transformed and structured the raw data properly
Designed the Airtable base manually (clean fields, correct data types, searchable structure)
Edited and optimized fields so the agent can search specific information, not just scan everything blindly
Once the data was ready, the workflow:
Searches Airtable based on the agent’s query
Uses IF conditions to intelligently differentiate between Bangladeshi users and foreign users
Pulls only the relevant, filtered data depending on the user type
Sends the exact information needed -no noise, no confusion
This wasn’t just automation. This was about decision-making with data.
What I love about n8n is how flexible it is when you combine:
AI agents
Datastores
Airtable logic
Conditional flows
You can literally build systems that behave like a mini backend + brain combined.
Every workflow like this makes automation feel more powerful and more real.
If you’re building with n8n, Airtable, or AI agents, you know how satisfying this feels. Let’s keep building.

01/01/2026

I just built a Learning Roadmap AI Agent that automatically creates a personalized learning path for users from a simple form submission.

Here’s how the system works behind the scenes (keeping it simple):
• User submits a form with their name, email, and learning goal
• The data is stored cleanly in Airtable
• A switch logic helps route the request properly
• The AI agent (Google Gemini) generates a complete learning roadmap
• The roadmap is then written back to Airtable in a dedicated “Roadmap” column
No manual work. No copy-paste. Just a smooth, end-to-end automation.

What I liked most while building this was realizing how powerful AI becomes when it’s combined with the right workflow, not just prompts.
This kind of system can be used for:
– education platforms
– onboarding users
– coaching programs
– or even personalized email delivery later on.

29/12/2025

Here’s how I designed my own Script Generator Agent step by step 👇
Workflow:
👉 Zapier Interfaces → AI by Zapier → Google Docs
Step 1: Zapier Interfaces (Frontend)
Instead of forms or chats everywhere, I used Zapier Interfaces as a clean input layer.
Users just enter:
✅ Topic
✅ Platform (YouTube / Reels / Ads, etc.)
✅ Tone & goal
Simple input. Clear intent.

Step 2: AI by Zapier (Brain)
This is where the real logic lives. I trained the AI with:
✅ Structured prompt logic
✅ Script framework (hook → body → CTA)
✅ Tone control (casual, professional, persuasive)
So the AI doesn’t just “write”, it thinks like a scriptwriter.

Step 3: Google Docs (Output)
Instead of dumping text in emails or chats, the final script is automatically generated into Google Docs. Why this matters:
✅ Easy editing
✅ Shareable
✅ Client-ready
✅ Scalable
Why I built it this way
❌ No manual copy-paste
❌ No prompt rewriting every time
✅ Reusable system
✅ Real automation, not just AI usage.

28/12/2025

I built an Email Automation system in zapier where

👉 someone sends me an email

👉 an AI agent understands the intent

👉 and sends a relevant reply automatically

Not auto-replies like “Thanks for reaching out” but context-aware replies, based on real information and data I’ve already provided to the AI.

What I focused on while building this:

✅Understanding email intent, not just keywords

✅Making replies sound human, not robotic

✅Ensuring the AI answers only from approved data

✅Keeping it scalable (so it doesn’t break when volume increases)

This kind of automation is super useful for:

✔️Support emails

✔️Lead inquiries

✔️FAQ-heavy inboxes

✔️Solo founders who can’t reply 24/7

The biggest lesson?

👉 Automation isn’t about replacing humans.

👉 It’s about removing repetitive work so humans can focus on important things.

If you’re exploring AI agents, email workflows, or smart automation or want to do repetative task for your business, let’s connect .

AutoTaranga

15/01/2026

14/01/2026

09/01/2026

08/01/2026

06/01/2026

06/01/2026

04/01/2026

04/01/2026

02/01/2026

01/01/2026

29/12/2025

28/12/2025

Address

Website

Alerts

Shortcuts

Share