Evolving December 10, 2025

Job Pics Field Workflow System

From 8:30pm backlog to real-time on-site photo pipelines

field workflows automation mobile ops systems
Telegram Bot API Google Drive API LLM Integration Email Automation

1. Context

This system was built for a small service business with field technicians who complete jobs throughout the day and need to send photos for social media and documentation.

Organic social media posting matters to the company, but consistency requires photos to arrive when jobs are completed, not hours later.

The original workflow relied on email. It wasn’t working.

2. The Problem

The workflow broke at the human edge.

After setting up Google Workspace for the team, I created a shared Drive and email address for job photos. Initially, it seemed like a clean solution.

Then I noticed the emails were not arriving until late at night, the next day, or not at all.

I spoke with the tech. He explained:

  • He doesn’t use email personally
  • Entering addresses, subjects, and attaching photos on mobile felt clunky, especially on site
  • He’d batch them at home in the evening when he had time

The problem wasn’t discipline. It was system design.

The tech felt bad about the delays. But that conversation revealed the real insight: the best solution would meet him where he already operates.

3. Constraints

  • Needed to be mobile-first
  • Had to work with minimal typing
  • Required instant visibility for the owner
  • Needed zero new tools for internal admin
  • Had to integrate with existing Google Workspace tools

4. The System I Designed

A Telegram-based photo logging pipeline that meets the tech where he already operates.

v1 Components (Deployed September 2025)

  • A private Telegram channel
  • A custom Telegram bot
  • Google Drive folder automations (date-based)
  • Caption-to-textfile saver
  • Real-time job folder organisation

How v1 Works

  1. The tech opens Telegram and takes photos directly from the channel
  2. Adds a quick caption and presses send
  3. The bot receives the message instantly
  4. Photos are saved into a Google Drive folder named with the current date
  5. Captions are stored as .txt files alongside the photos
  6. The business owner sees new jobs immediately, without chasing

v2 Enhancement (Deployed December 2025)

The original system delivered photos in real time, but I still needed to:

  • Open Google Drive to check for new uploads
  • Download photos manually
  • Rewrite brief captions into social media-ready copy

The v2 enhancement automated the final mile:

  • Captions are sent to an LLM for rewriting (with style guidance from previous posts)
  • Original caption + LLM suggestions + photos are bundled into a single email to marketing
  • Ready for social posting in minutes, with no Drive downloads required

The LLM suggestions are still being refined, but the bundling alone saves significant time.

5. Before vs After

BeforeAfter
Photos sent late at nightReal-time on-site uploads from day one
High friction email workflowOne-tap mobile workflow
Lost or forgotten photosConsistent daily uploads
Owner checking Drive manuallyAutomated email with photos + captions
Caption rewriting overheadLLM-assisted drafts + original text

6. Stack & Architecture

v1 (September 2025)

  • Telegram Bot API for low-friction mobile UX
  • Google Drive API for folder creation + file organisation
  • Google Scripts for webhook-to-storage logic
  • Simple date-based foldering system for clarity and batching

v2 (December 2025)

  • LLM integration for caption rewriting with style guidance
  • Email automation to bundle original captions, LLM suggestions, and photos
  • Archived folder system for posted content organisation

This is an example of designing a system around the human constraints, not the ideal scenario.

7. Impact

  • The tech now uploads photos while still on site
  • Increased frequency and reliability of documentation
  • No more “8:30pm admin time”
  • Real-time visibility into completed jobs
  • Time saved redirected to higher-value work: digital calendar implementation, SEO articles, client follow-up SMS
  • Maintained social media consistency without additional admin overhead

8. What I’d Build Next (v3)

  • Refine LLM prompt engineering for better caption suggestions
  • Add automatic social media post scheduling
  • Tagging or recognition for job types
  • Auto-classify the best photo in each batch for featured posts