AI Systems Engineer
Saadi Mughal
AI Automation &
Agentic Systems Engineer
Building
AI Employees CRM Automation Systems WhatsApp Workflows AI Customer Support Business Process Automation Cybersecurity AI
20+
Automation Workflows Built
1.2K+
LinkedIn Followers
85%
Manual Overhead Reduced
30K+
B2B Professionals Reached
The Engineer Behind
the Systems
I design AI-powered systems that help businesses automate repetitive processes, improve operational efficiency, and create scalable workflows using n8n, AI agents, and modern backend tools.

Based in Multan, Pakistan, I work with international clients across Europe — building production-ready systems that run 24/7 without manual intervention. My focus is on agentic architectures, RAG systems, and multi-step intelligent workflows that replace entire human workflows with reliable automation.

I don't just build workflows — I build AI employees.
Location
Multan, Punjab, Pakistan
Clients
International — Europe & beyond
Education
ICS Computer Science + Self-Directed AI Learning
Leadership
PSEB Soft Skills Program — Team Leader
Status
Open to Freelance & Remote
What I Build
🤖
AI Employee Systems
Autonomous AI agents that handle tasks, make decisions, and execute multi-step workflows — replacing manual processes entirely.
📞
AI Receptionists
24/7 voice and chat AI that handles inbound inquiries, books appointments, and routes complex cases to humans automatically.
⚙️
CRM Automation
End-to-end CRM pipelines — auto-sync contacts, trigger follow-ups, score leads, and keep your data clean without manual effort.
💬
WhatsApp Automation
AI-powered WhatsApp bots with RAG memory, lead qualification, and CRM sync via WhatsApp Business Cloud API.
🛡️
AI Cybersecurity Systems
Automated threat intelligence pipelines — phishing triage, malware scoring, and real-time security alerting without manual analysis.
🏗️
Internal Business Systems
Custom operational automation — reporting, data pipelines, notification systems, and workflow orchestration for internal teams.
Case Studies
01 Office Automation
AI Office System
A complete AI-powered office that manages, communicates, and operates 24/7 — without human intervention.
n8n OpenAI Gmail API WhatsApp API Google Sheets Slack
+
System Architecture
AI Office System Architecture
n8n Workflow — Live Editor
AI Office System n8n Workflow
The Problem
Businesses were losing leads and customer inquiries because office operations required constant human attention. Emails went unanswered, form submissions sat unprocessed, and there was no unified system to handle multi-channel communication at scale.
The Solution
Built a fully automated AI office that captures inputs from website forms, email inquiries, and WhatsApp simultaneously. An OpenAI-powered AI Agent classifies intent, generates responses, and routes to the correct action path — all without human involvement.
System Architecture
01
Input Collection
Gmail Trigger + Form Submission merged into unified data stream
02
AI Processing
OpenAI Agent analyzes intent, generates contextual response
03
Smart Routing
Escalation filter: priority cases get immediate multi-channel alert
04
Action & Follow-up
Gmail + Slack notifications, Google Sheets CRM logging
✓ 100% Automated Response ✓ Multi-channel (Email + WhatsApp + Slack) ✓ Zero Human Intervention ✓ 24/7 Operation ✓ Saves 80% Manual Work
02 Cybersecurity
AI Cyber Security Co-Pilot
Automated threat intelligence pipeline — scans URLs and files, scores risk, and sends security reports in real-time.
n8n VirusTotal API Grok AI Gmail Google Sheets
+
n8n Workflow — Threat Analysis Pipeline
AI Cyber Security Co-Pilot Workflow
The Problem
Security teams manually checked suspicious URLs and files one by one — a slow, error-prone process that left organizations exposed. Threat intelligence took hours instead of seconds.
The Solution
An automated security pipeline that ingests IOCs (Indicators of Compromise) via webhook, scans them against VirusTotal's database, extracts threat data, calculates a risk score using AI, generates a structured report, and sends alerts — all within seconds.
Pipeline Phases
01
Data Entry
Webhook receives IOC → Data Normalize → VirusTotal API scan → Wait
02
Intelligence Extraction
Get Results → VT Data Extract → Grok API analysis → Structured response
03
Risk Scoring
AI calculates risk score, generates full threat report, logs to Google Sheets
04
Alerting
Email security report sent via Gmail with full analysis attached
✓ Real-time Threat Detection ✓ AI Risk Scoring ✓ Automated Security Reports ✓ VirusTotal Integration
03 Cybersecurity
PhishGuard Pro
Automated phishing email triage and threat intelligence pipeline with Claude AI analysis and SIEM integration.
n8n Claude AI VirusTotal Gmail PDF Reports SIEM Alert
+
n8n Workflow — 6-Phase Phishing Pipeline
PhishGuard Pro Workflow
The Problem
SOC teams were overwhelmed with phishing email volume. Manual triage was slow, inconsistent, and caused alert fatigue. Attachments and URLs went unchecked, and threat reporting took hours per incident.
The Solution
A 6-phase automated pipeline that ingests phishing emails, simultaneously analyzes headers, URLs, and attachments in parallel, feeds all intelligence into Claude AI for threat analysis, generates PDF reports, and fires SIEM alerts for critical threats.
6-Phase Pipeline
01
Email Ingestion & Keyword Scan
Gmail monitor + keyword detection + email parser
02
Parallel Intelligence
Header analysis, URL scanning (VirusTotal), Attachment analysis
03
AI Threat Analysis
Claude AI merges all signals → threat classification → report parsing
04
Reporting & Notifications
HTML report → PDF conversion → trust router → multi-channel notify
05
SIEM Alert
Critical threats trigger automated SIEM integration
✓ Parallel Threat Analysis ✓ Claude AI Integration ✓ Auto PDF Reports ✓ SIEM Integration ✓ Zero Manual Triage
04 Customer Support
SMB AI Support Router
Intelligent support ticket routing with sentiment analysis, department classification, and human-in-the-loop validation.
n8n OpenAI Google Sheets Telegram Human-in-Loop
+
n8n Workflow — Support Router with Human Validation
SMB AI Support Router Workflow
The Problem
SMBs struggle with support tickets getting routed to the wrong department, causing delays and frustrated customers. AI-only routing risked misclassification with no human oversight. Manual routing was too slow.
The Solution
A hybrid AI + human system that performs sentiment analysis, classifies tickets to the correct department (Sales, Technical, Billing, General), calculates priority, and — for borderline cases — asks a human to approve or correct the AI's decision before proceeding.
Workflow Logic
01
Ticket Intake
Support Ticket Trigger → OpenAI Sentiment Analysis
02
Department Classification
AI classifies → Sales / Technical / Billing / General
03
Priority Scoring
Calculate priority → check if human review required
04
Human Validation
Chat approval: confirm sentiment + department, or override
05
Resolution
Generate AI response → log to Support Tickets Table → Create task
✓ Sentiment Analysis ✓ Smart Department Routing ✓ Human-in-the-Loop ✓ AI Response Generation
05 CRM & Sales
CRM Agent with Vector Memory
AI-powered CRM agent using Supabase Vector Store for semantic search — retrieves customer context and sends intelligent, personalized responses.
n8n OpenAI Supabase Vector Store Webhook JavaScript
+
n8n Workflow — CRM Agent with Vector Store
CRM Agent Workflow
The Problem
CRM systems store vast customer data but agents can't retrieve the right context quickly. Support reps wasted time searching for customer history, leading to generic, unhelpful responses that damaged customer relationships.
The Solution
A RAG-powered CRM agent that stores all customer information in a Supabase Vector Store. When a customer inquiry arrives via webhook, the AI Agent performs semantic search, retrieves relevant context, and generates a personalized, accurate response — instantly.
Architecture
01
Getting Messages
Webhook (POST) receives customer inquiry
02
Vector Search
Supabase Vector Store + OpenAI Embeddings for semantic retrieval
03
AI Agent Processing
OpenAI Chat Model + Simple Memory generates contextual response
04
Response Delivery
JavaScript formatting → Respond to Webhook
✓ Semantic Vector Search ✓ Supabase RAG ✓ Personalized Responses ✓ Conversation Memory
06 Outreach Automation
Cold Email Automation Pipeline
End-to-end cold outreach engine — extracts contacts, deduplicates, sequences sends with AI-written personalized emails.
n8n OpenAI Gmail Webhook Airtable
+
n8n Workflow — Cold Email Pipeline
ColdEmail Pipeline Workflow
The Problem
Sales teams spent hours manually writing cold emails, managing follow-up sequences, and tracking send status. Personalization was inconsistent and leads frequently slipped through the cracks without proper follow-up.
The Solution
A fully automated pipeline that extracts email contacts, checks for duplicates, stages them in a sequence database, then uses an OpenAI AI Agent to write personalized cold emails and sends them via Gmail — with intelligent timing and wait logic between sequences.
Pipeline Stages
01
Email Extraction
Webhook → Edit Fields → Run Actor & get dataset → Loop Over Items
02
Saving & Deduplication
Edit Fields → Check if row exists → Loop → Insert row → Loop
03
Send Sequencing
Switch rules → Loop → Wait timer → Loop Over Items
04
AI Writing & Sending
AI Agent (OpenAI) writes personalized email → Gmail send
✓ Automated Personalization ✓ Duplicate Prevention ✓ Sequence Timing ✓ AI-Written Emails
07 RAG System
RAG Knowledge Agent for MIT
Retrieval-Augmented Generation system with OpenAI embeddings — answers complex queries from a custom knowledge base with high accuracy.
n8n OpenAI Embeddings Simple Memory Webhook JavaScript
+
n8n Workflow — RAG Agent Architecture
RAG for MIT Workflow
The Problem
Organizations with large knowledge bases faced high query response costs and slow retrieval. Generic AI responses lacked domain-specific accuracy — staff wasted time searching internal documents manually.
The Solution
A RAG pipeline that converts organizational knowledge into OpenAI vector embeddings, enabling semantic search. An AI Agent retrieves the most relevant context and generates accurate, grounded answers — with conversation memory for follow-up questions.
Architecture
01
Query Intake
Webhook (POST) receives user question
02
Embedding Search
OpenAI Embeddings perform semantic similarity search on knowledge base
03
AI Agent Response
OpenAI Chat Model + Simple Memory generates grounded answer
04
Output Processing
JavaScript formatting → Respond to Webhook
✓ Semantic Vector Search ✓ Grounded AI Answers ✓ Conversation Memory ✓ Custom Knowledge Base
08 WhatsApp AI
WhatsApp AI Agent with Booking
Intelligent WhatsApp bot with availability checking, ticket creation, and appointment booking — a complete AI receptionist on WhatsApp.
n8n OpenAI WhatsApp API Google Calendar Google Sheets
+
n8n Workflow — WhatsApp AI Agent
WhatsApp AI Agent Workflow
The Problem
Businesses using WhatsApp for customer communication had no intelligent automation — every message required manual response. Appointment booking was done manually, tickets were created by hand, and availability checking wasted staff time.
The Solution
A WhatsApp-native AI agent that intercepts every message, runs it through a condition check, then uses an OpenAI-powered agent with three tools: check_availability (Google Calendar), create_Ticket (Google Sheets), and book_appointment (Google Calendar).
Agent Tools
WhatsApp Trigger
Receives incoming message, routes through condition check
AI Agent (OpenAI)
Chat Model + Simple Memory for context-aware conversations
🔧
check_availability
Queries Google Calendar for open slots
🔧
create_Ticket
Appends support ticket to Google Sheets
🔧
book_appointment
Creates event in Google Calendar, sends confirmation
✓ 3 AI Tools Integrated ✓ WhatsApp Native ✓ Auto Booking ✓ Conversation Memory
09 Lead Generation
AI Lead Generation & Outreach System
10-step automated sales machine — from lead source scraping to personalized multi-channel outreach, qualification, and CRM update.
n8n Apollo API LinkedIn Scraper OpenAI Gmail WhatsApp Calendly Airtable
+
n8n Workflow — 10-Step Lead Generation System
AI Lead Generation System Workflow
The Problem
Sales teams were manually sourcing leads from LinkedIn and Apollo, cleaning data by hand, qualifying leads one by one, and sending generic outreach messages. The entire process took days and produced poor conversion rates.
The Solution
A 10-step automated pipeline that sources leads from Apollo, Google Maps, and LinkedIn, deduplicates and cleans data, qualifies via AI (HOT/WARM/COLD), sends personalized multi-channel messages, detects replies with AI, books appointments, updates CRM, and runs analytics — entirely automated.
10-Step Pipeline
01
Lead Source
Apollo, Google Maps Scraper, LinkedIn Scraper → Merge → Add to Sheet
02
Data Cleaning & Deduplication
Set Fields → IP Dedup → Discard Duplicates
03
AI Lead Qualification
OpenAI → Set Qualification: HOT / WARM / COLD / Revision
04
Personalized Message Generator
OpenAI generates Hi [Name], I noticed [Company] is working on [something]...
05
Multi-Channel Outreach
Gmail + WhatsApp API + LinkedIn message simultaneously
06
Reply Detection AI
Gmail Watch → OpenAI classifies: Interested / Not Interested
07
Appointment Booking
Gmail send Calendly link → Calendly Create Invite → Send Invite
08
CRM Update
Airtable Create/Update → Update Lead Notes
✓ 3-Source Lead Scraping ✓ AI Qualification (HOT/WARM/COLD) ✓ Multi-Channel Outreach ✓ Reply Detection AI ✓ Auto Appointment Booking ✓ CRM Auto-Update
System Design Philosophy
Stateless Orchestration
Each workflow node operates independently with no hidden state — making systems predictable, debuggable, and horizontally scalable.
Centralized Operational State
All system state lives in a single source of truth (Supabase/PostgreSQL) — eliminating data inconsistency across distributed workflow nodes.
AI + Human Handoff
Every AI system has defined escalation paths — when confidence is low or context requires human judgment, the system routes intelligently.
Deterministic Workflows
Business logic is encoded in deterministic steps. AI handles reasoning; workflows handle execution — keeping systems reliable and auditable.
Supabase + Edge Functions
Real-time database triggers + serverless edge functions enable reactive systems that respond instantly to data changes without polling.
Reliable State Management
Idempotent operations and transaction logging ensure workflows recover gracefully from failures — no lost data, no duplicate executions.
Tools & Technologies
Automation & Orchestration
n8nMake.comZapierWebhooksREST APIs
AI & Language Models
OpenAI GPT-4Claude AIGrok AIRAG SystemsPrompt Engineering
Databases & Backend
SupabasePostgreSQLVector DBAirtableGitHub
Communication & Voice
WhatsApp APITwilioElevenLabsSlackTelegramGmail API
Security & Intelligence
VirusTotal APISIEM IntegrationThreat Intelligence
Business & Payments
StripeCalendlyGoogle SheetsGoogle Calendar
Languages
JavaScriptPythonJSONSQL
Building in Public
Sharing the System
Behind the Systems
I document real AI automation builds, architecture decisions, and agentic system design — reaching thousands of B2B professionals every week with technical case studies.
Follow on LinkedIn
Content Topics
AI Automation Case Studies System Design Agentic AI n8n Builds RAG Architecture
90K+
Impressions
30K+
Professionals Reached
1.2K+
Followers
2mo
To Achieve This
Let's Build Smarter Systems Together
Open to freelance projects, remote contracts, and long-term partnerships with teams who want to integrate serious AI automation into their operations.
Book a Free Call