Sitemap

🧠 How I Built an AI Assistant That Handles My Instagram DMs and Customer Photos for CanvasMagnet

2 min readMay 23, 2025

As a solo founder running CanvasMagnet, I used to spend hours replying to DMs, following up with customers, and prepping images for print.

Eventually, I asked myself:
β€œWhy am I doing this manually when AI can do it better and faster?”

So I built my own AI system that automates the entire customer communication workflow.

πŸ”„ What My AI Assistant Does

βœ… Automatically reads Instagram DMs
βœ… Analyzes full chat history using GPT
βœ… Decides what to do: follow-up, remind, ignore, or confirm order
βœ… Sends context-aware replies that feel human
βœ… Logs every message into a local database for training or fine-tuning later
βœ… Waits random intervals to avoid spam detection

The assistant is smart enough to recognize:

  • If a customer says β€œI’ll message later” β†’ it pauses
  • If they say β€œHow much for 9 photos?” β†’ it sends pricing
  • If they send their address and photos β†’ it confirms the order
  • and so on..

πŸ–Ό Meet SoftCropper: My Photo Automation Tool

Besides DMs, I also had to prepare customer photos for printing. It was time-consuming and error-prone.

So I built SoftCropper, a Python tool that:

  • Resizes photos to square (5x5cm)
  • Adds blurred borders
  • Generates clean, print-ready images

It saves 15–30 minutes per order, and I don’t have to touch Photoshop.

You can install it via PyPI:

pip install softcropper

PyPI: SoftCropper

πŸš€ What’s Next?

  • Trainf and feed message logs into Local models
  • Build a simple CRM-like dashboard
  • Add support for WhatsApp automation
  • Let the bot suggest upsells and new package offers

πŸ‘‹ Want to Try It or See a Demo?

Message me at:
πŸ“§ khaledalam.net@gmail.com
πŸ“± WhatsApp: +971 58 293 6628
🌐 khaledalam.net

--

--

Khaled Alam
Khaled Alam

Written by Khaled Alam

Senior Software Engineer, Lead

No responses yet