# The Clients

## Primary Client

### Imona Logistics

Imona Logistics is a U.S.-based trucking and freight company managing an active fleet of 19 professional drivers. Operating in the freight brokerage market, the company receives broker load requests via email with PDF attachments.

Between May 2025 and August 2026, the automated system processed 9,851 load PDF files for Imona, eliminating 80+ hours per month (1,641 hours total) of tedious manual data entry.

- **80+ hrs/mo Saved Monthly**  
- **9,851 PDFs Processed**  
- **1,641 hrs Total Saved**  
- **19 Active Drivers**

---

## Referral Client

### Avesto LLC

Avesto LLC is a sibling trucking operation run by the brother of Imona Logistics' CEO. After seeing Imona eliminate manual dispatch bottlenecks, Avesto's leadership requested the identical automation framework.

Between May 2025 and August 2026, the system processed 9,952 load PDF files for Avesto, reclaiming 80+ hours per month (1,658 hours total) with zero marketing spend.

- **80+ hrs/mo Saved Monthly**  
- **9,952 PDFs Processed**  
- **1,658 hrs Total Saved**

# The Challenge

Before automation, Imona's dispatch workflow involved a five-step manual routine repeated for each of the 20–25 loads received daily.

**~10 minutes per load** for manual PDF data extraction, adding up to **80–100 hours** of administrative work each month.

1. **Open every email & PDF manually**  
   Each broker sends PDFs in different layouts, no standardization, all read by hand.
2. **Copy & format key load data**  
   Pickup/delivery locations, rates, mileage, all extracted manually and reformatted for internal use.
3. **Send to Telegram dispatch group**  
   Formatted load info typed or pasted manually into a Telegram group where drivers see available jobs.
4. **Log all data into Google Sheets**  
   Every load recorded manually, creating a second data-entry step for the same information.
5. **Monitor Telegram & update the sheet**  
   When a driver accepted a job via chat, someone had to catch it and manually update the spreadsheet.

### Key Operational Challenges

- **Data Entry Errors**  
  Copying details across tools manually led to occasional errors in rates, locations, or driver assignments.
- **Dispatch Response Times**  
  Slower response times to broker emails made it harder to secure preferred loads quickly.
- **Tracking Driver Acceptances**  
  Checking Telegram chats manually meant driver responses had to be logged into the spreadsheet by hand.
- **Scaling Constraints**  
  Handling higher load volumes required additional time and manual effort from the dispatch team.
- **Administrative Overhead**  
  Dispatchers spent a large portion of their workday on repetitive data entry instead of logistics planning.

# The Solution

Flowbetta designed a **3-layer automation system** using Google Apps Script to connect existing tools (Gmail, Telegram, Google Sheets, Google Drive)—without requiring new software subscriptions or changing team habits.

**Gemini AI** was integrated to read incoming broker PDFs in varying layouts and extract structured load data into a consistent format.

1. ### Email → Telegram & Drive
   The system checks Gmail every 10 minutes for new broker load emails. When received, it passes the PDF attachment to Gemini AI, which extracts key details (origin, destination, rate, mileage, pickup windows) and posts them to the dispatch Telegram group while archiving the PDF to Google Drive.

Gmail → PDF Attachment → Gemini AI → Telegram Group → Google Drive

2. ### Real-Time Data Logging  
   Simultaneously, the extracted load data is populated into a dedicated Google Sheet. Each load creates a new structured row, giving the team a centralized log of all incoming requests without manual copy-pasting.

Gemini Extract → Google Sheets → Structured Row → Central Log

3. ### Automated Driver Assignment  
   The system monitors driver Telegram chats every 30 minutes. When a driver accepts a load, the script identifies the message and updates the corresponding row in the Google Sheet, marking the load as assigned.

Telegram Chat → Driver Acceptance → Script Detects → Sheet Updated

## Full Tech Stack

- Gmail API
- Gemini AI
- Telegram Bot API
- Google Apps Script
- Google Sheets
- Google Drive

# The Impact

**📅 Live Production Verification:** May 2025 – August 2026 (16 Months)

### Imona Logistics
- **Primary Fleet**
  - **9,851 Load PDFs Processed**
  - **1,641+ hrs saved**
  - **98,510 manual minutes eliminated (~102.5 hrs saved every month)**

### Avesto LLC
- **Referral Fleet**
  - **9,952 Load PDFs Processed**
  - **1,658+ hrs saved**
  - **99,520 manual minutes eliminated (~103.6 hrs saved every month)**

### Monthly Savings
- **160+ hrs/mo**
- **Combined Load PDFs: 19,803 files**
- **Total Time Saved: 3,299 hours**
- **Productivity Equivalent: ~412 workdays**

### The Efficiency Calculation: 10 Minutes Saved Per Load PDF
Prior to Flowbetta's Gemini AI automation, a dispatcher required approximately **10 minutes per PDF load document** to open the email, read the unstandardized rate sheet, extract origin/destination/rates, retype the summary into Telegram, and log the row into Google Sheets.

- **Imona Logistics Calculation:** 9,851 files × 10 min = 98,510 min = **1,641.8 hours**
- **Avesto LLC Calculation:** 9,952 files × 10 min = 99,520 min = **1,658.6 hours**

⚡ **Instant Load Broadcast**  
What previously took ~10 minutes of manual handling is parsed and broadcast to Telegram in under 1 minute.

🎯 **Flawless Data Extraction**  
Over 19,800 PDFs parsed with Gemini AI across hundreds of broker document formats without data loss.

📈 **Zero Overhead Scaling**  
Both fleets doubled processing capacity over 16 months without adding a single administrative hire.

# Behind the Build

The second deployment came about when Imona Logistics' CEO introduced Flowbetta to his brother's company after seeing positive results in his own operations.

"Practical results build confidence."

Expanding the workflow to a second fleet confirmed that a clean automation structure could be adapted easily to similar logistics routines.

Open communication during initial testing—handling edge-case PDF layouts and API quota limits—helped maintain clarity and steady progress throughout setup.

### Technical Adjustments
- **Varied PDF Layouts**  
  Broker PDFs came in different formats. Gemini AI prompts were refined to extract data accurately across layouts.
- **Gmail API Quota Limits**  
  Polling intervals and request batching were adjusted to stay within standard Google Workspace usage limits.
- **API Retry Logic**  
  Added retry logic and error logging to ensure reliable data transfer across Telegram and Drive APIs.
- **Telegram Chat Parsing**  
  Applied clear message parsing rules to confirm driver acceptances without misinterpreting casual chat messages.

# Key Lessons

### Understand the Workflow First
Mapping out manual steps thoroughly before writing code ensured the system fit how dispatchers actually work.

### Clear Communication Builds Trust
Communicating openly during testing and resolving early edge cases promptly kept project progress smooth.

### Work Within Existing Tools
Using Gmail, Sheets, Drive, and Telegram minimized setup overhead and helped the team adopt the system quickly.

### Client Referrals Drive Growth
Delivering practical, reliable value for the primary client led directly to a word-of-mouth referral.