Scrolling LinkedIn jobs. Saw: "Hiring Data Entry Clerk - Process 100 shipping manifests weekly." I didn't apply. I messaged the hiring manager: "What if those manifests processed themselves?" THE SITUATION: Small logistics company. Posted job 3 weeks ago. No good candidates. Needed someone to manually extract data from BOLs and manifests into their TMS system. Annual cost if they hired: $32,000 salary plus benefits. My message: "I can automate manifest processing. Gmail receives documents, extracts all data automatically, posts to your TMS. Want to see a 10-minute demo?" Reply came 4 hours later: "Very interested. Can we talk tomorrow?" THE DEMO: Screen share. Opened n8n. Built workflow live: Gmail trigger → Parse Document → Extract structured data (shipper, consignee, BOL number, weights, destinations) → HTTP Request to their TMS API Showed it processing 3 real manifests. Data appeared in their system. Build time during call: 35 minutes. They watched the whole thing. THE CLOSE: "This would save us the hiring process, training time, and $32K annually. What do you charge?" My quote: $1,600 setup + $100 monthly maintenance. They signed same day. THE LESSON: Job postings aren't just jobs. They're automation opportunities. "Hiring data entry" means "doing manual document processing." That's literally what automation solves. WHERE TO LOOK: LinkedIn: Search "data entry" + "hiring" in your city Indeed: Filter for admin/data entry roles Company career pages: Look at their open positions WHAT TO SAY: Don't: "I'm an automation expert looking for work" Do: "Saw your posting for [job title]. What if [their manual task] happened automatically?" THE WORKFLOW I BUILT: Gmail → PDF Vector Parse Document → Extract Structured with schema → Format data → HTTP to TMS → Archive email Template pattern in n8n Build time: 6 hours total (including testing and refinements) Their savings: $32K annually My fee: $1,600 one-time

Property management company. 180 inspection reports annually. Manual review consuming 2 hours per report. Built automated inspection processor. Eliminated manual analysis completely. THE CLIENT'S MANUAL PROCESS: Every inspection report requiring manual handling. Property manager reading entire 20-40 page document. Manually noting issues across different locations. Opening spreadsheet typing each issue - location, description, estimated cost. Attempting severity categorization subjectively. Manually flagging safety hazards. Calculating total repair costs. Determining property risk level without systematic criteria. Creating summary for ownership decision. 2 hours per report. 180 reports annually. 360 hours total. Quality audit revealed problems - 23 reports had critical safety hazards not flagged. 45 reports with major system failures categorized incorrectly. 67 reports with inaccurate cost totals from manual math errors. THE AUTOMATION SOLUTION: 8-node workflow with risk-based notification routing: Google Drive Trigger → Download → Prepare Binary → PDF Vector Extract → Analyze & Prioritize → Log Register → Generate Summary → Risk Router → [Critical Alert] OR [Normal Notification] Monitors shared Drive folder. New inspection report uploaded triggers automatic processing. EXTRACTION DETAILS: Property - address, type, age, overall condition Inspection - date, inspector name, company Issues - location, category, description, severity, cost, urgency, safety hazard Major systems - roof, HVAC, plumbing, electrical, foundation status PRIORITY SCORING: Base severity score (Critical=100, High=75, Medium=50, Low=25) Add 25 points if safety hazard present Add 15 points if urgency immediate Sort all issues by priority descending Top 3 issues flagged for immediate attention RISK LEVEL LOGIC: Critical issues OR safety hazards = Critical risk 3+ high severity OR $10,000+ repairs = High risk 1-2 high issues OR 5+ medium = Medium risk Otherwise Low risk Systematic criteria applied consistently.

Built with: n8n · WhatsApp Cloud API · Google Gemini · AI Agents Project Overview To strengthen my hands-on skills in automation and conversational AI, I built an AI-powered WhatsApp bot that can intelligently handle text, voice, and image messages—all within a single automated workflow. This project demonstrates how WhatsApp can be transformed from a simple messaging channel into a smart, multi-modal assistant capable of understanding user intent and responding contextually. Key Use Cases - Customer Support Automation - WhatsApp Virtual Assistants - Voice-to-Text Chatbots - Image-Based Queries (Product, Documents, etc.) - Lead Qualification & FAQ Bots Learning Outcomes ⚡ Built multi-modal WhatsApp automation (text, voice, image) ⚡ Hands-on experience with AI agents and message routing ⚡ Improved workflow design using n8n best practices ⚡ Practical understanding of real-world chatbot architecture

University admissions. 1,200 transcripts annually. Manual entry consuming 900 hours. Built automated processor. Eliminated transcription. THE CLIENT'S MANUAL PROCESS: Every transcript requiring manual processing. Coordinator downloads PDF attachment from email. Opens admissions system. Types student name, student ID, email address manually. Enters institution details - name, address, registrar information. Opens course entry screen. Types each course individually - course code, course name, credit hours, letter grade, semester completed. 15-30 courses per transcript average. Every course manually transcribed from PDF. Sums credits attempted and earned manually. Verifies GPA calculation against transcript. Notes academic honors if present. Determines transcript type - official or unofficial. Checks for registrar signature presence. Flags verification requirement if official transcript with student ID. 45 minutes per transcript. 1,200 transcripts annually. 900 hours total consumed. Audit revealed problems - 178 course entry errors from typing mistakes. 89 GPA calculation mistakes from manual math. 234 missing academic honors notation from overlooked entries. 67 transcripts incorrectly marked verified. 123 flagged for verification never processed. THE AUTOMATION SOLUTION: 7-node workflow with verification routing: ``` Gmail Trigger → Get Email → Prepare Binary → PDF Vector Extract → Analyze Transcript → Log Database → Verification Router → [Check] OR [Unverified] ``` Monitors inbox. Transcript arrives. Triggers processing. EXTRACTION: Student - name, ID, DOB, email Institution - name, address, registrar information Degree - program, major, minor, enrollment/graduation dates Courses - complete list with codes, names, credits, grades, semesters Academic - credits attempted/earned, GPA values, honors Metadata - transcript type, date, registrar signature ANALYSIS: Total courses processed Completion rate (earned/attempted percentage) GPA categorization (Exceptional/Excellent/Good/Fair)

Property management company. 240 new leases annually. Manual data entry consuming 3 hours per lease. Built automated lease processor. Eliminated transcription completely. THE CLIENT'S MANUAL PROCESS: Every signed lease requiring manual handling. Property manager downloading PDF. Opening CRM typing tenant details. Opening accounting entering financial terms. Opening calendar adding lease dates. Composing welcome email manually. Copying information from lease document. Calculating total upfront costs. Including move-in checklist, payment instructions, property details. 3 hours per lease. 240 leases annually. 720 hours total. Data audit revealed problems - 34 leases with incorrect rent in accounting. 67 missing renewal reminders. 89 tenants never received welcome emails. 19.4% error rate overall. THE AUTOMATION SOLUTION: 8-node workflow with conditional welcome email routing: Google Drive Trigger → Download → Prepare Binary → PDF Vector Extract → Analyze Lease → Log Database → Email Check → [Send Welcome] OR [Notify Logged] Monitors shared Drive folder. New lease uploaded triggers automatic processing. EXTRACTION DETAILS: Tenant information - name, email, phone, employment status Property details - address, unit number, parking allocation Financial terms - monthly rent, security deposit, pet deposit, rent due day Lease duration - start date, end date, term length in months Special provisions - pets allowed, utilities included, early termination clause CONDITIONAL WELCOME EMAIL: If lease upcoming within 30 days AND tenant email present - send personalized welcome email automatically including move-in checklist, payment details with calculated total, utility setup instructions, parking and property information, contact details. Otherwise - log to database, schedule for later or flag for manual contact if no email. DEPLOYMENT METRICS: 12 months operation. 240 leases processed automatically. Manual data entry eliminated - zero hours spent transcribing lease details. Zero typos in financial records. Zero discrepancies between lease documents and system records.