Scope of Work Spreadsheet: V4

From Walkthrough to AI: How I Automated Property Photo Labeling with a Web App + Make.com

Importance of this Build

If you manage real estate walkthroughs, construction projects, or rehab evaluations, you know how time-consuming it is to sort through 50+ photos after every site visit. I built a custom automation that turns property walkthrough photos into structured, labeled data — using a Node.js web app, Google Drive, and Make.com connected to ChatGPT Vision.

In this post, I’ll break down how I built the system — including the frontend, backend, and the Make.com AI automation that labels each photo.

Previous build:

I’ve documented my Scope of Work Spreadsheet for now 2 years and how it’s evolved after doing 50+ renovation projects.

The data i’ve collected along the way is nothing short of a miracle. It is pure GOLD!

Here’s my latest post on V3 of this build…

This build I just wanted to take things to a slightly higher level. There’s levels to these builds and each step I get closer to the most efficient version I have built in my head.

Firstly this build included the following updates

1. Manually coded and hosted the webapp using Node.JS instead of relying on Google App Script

   • The reason I did this is because I wanted more flexibility on the server side long term

2. Ability to add more photos

   • V3 was able to add maybe 20 photos max before it would inevitably crash which made things less efficient in the end because I would need to only upload like 10 photos and then wait until the automation was complete before going into the folder link and uploading the rest (inconvenient!!!)

3. Using ChatGPT to label the photos

   • This allows us to see in the folder what the photos are just by the names

   • This data will help build our vector database

   • Labeled photos down the road can be used in the place of manually entered data in the webform which means that the next build will remove the hardcoded data in the webform and will rely solely on the images to push the necessary data into the spreadsheet template

   • Photo labels can be used to create a report

---

Part 1: The Web App — Capturing Walkthrough Photos + Form Data

🛠 Tech Stack:

• Node.js + Express.js

• Google Drive API

• Google Sheets API

• Multer (for file uploads)

🧩 What the App Does:

• Hosts a mobile-friendly webform for in-field use

• Allows uploading of 50–100 photos

• Collects property details (unit breakdown, condition, etc.)

• On submission:

  • Creates a new subfolder in a Shared Drive (named after the address)

  • Uploads photos to that folder

  • Pushes form data to a Google Sheet

  • Copies a Scope of Work template spreadsheet

  • Returns a folder link for quick access

📄 Key Backend Logic:

js

CopyEdit

await drive.files.create({ name: file.originalname, parents: [folderId], ... }); await sheet.addRow(formData); await drive.files.copy({ fileId: TEMPLATE_ID, parents: [DEST_FOLDER_ID], ... });

The .txt file used for form notes was removed in the final version — keeping the Drive folder clean and automation-ready.

---

Part 2: The Make.com Automation — AI-Labeling the Photos

Once the walkthrough photos land in Drive, the next challenge is organizing them.

🧠 Enter AI

I used Make.com + OpenAI Vision to analyze each photo and rename it with a descriptive label like kitchen.jpg, bathroom_sink.jpg, electrical_panel.jpg.

---

🔁 Scenario Breakdown:

Module 1: Google Drive > Watch Files

• Watches the Walkthrough Photos parent folder

• Triggers when a new folder (e.g., "123 Main St") is added

Module 2: Google Drive > List Files in Folder

• Gets all photos inside the new subfolder

Module 3: Tools > Array Aggregator

• Combines all file bundles into a single array

Module 4: Tools > Iterator

• Loops through each photo

Module 5: Google Drive > Download File

• Pulls the actual image file from Drive

Module 6: OpenAI > GPT-4 Vision

• Sends the photo to ChatGPT with this prompt:

“You are labeling property walkthrough photos. Return a one- or two-word label like kitchen, bathroom_sink, water_heater, or exterior_front.”

Module 7: Google Drive > Update File

• Renames the photo in Drive using the AI-generated label

---

🚀 Final Result:

Every walkthrough folder automatically turns into a structured, searchable photo library with labeled images — no more guessing what “IMG_4092.jpg” is.

---

💡 Use Cases

• Construction managers labeling pre-rehab photos

• Property managers preparing leasing packages

• Real estate investors building reports from site visits

---

📎 Wrap-Up

This automation is a huge game changer for organizing files seamlessly and removed a few bottlenecks we had with the previous builder. As with any internal app or automation workflow process you’ll constantly need to update and revise it and continuously improve it to keep up with the demands of a growing company.