You've moved. The boxes are stacked in the garage, in the pantry, at your mother-in-law's. You're looking for the coffee maker. Which box is it in? Instead of opening every box, reading every marker-scrawled list, digging for 20 minutes, a searchable index gives you the answer by typing one word.

But for the index to exist, every box first has to be documented. That's where KontexVoX comes in: you photograph the open box before closing it, you say "Box 14: coffee maker, filters, blue mugs, Italian cookbook, dish towels." The description is engraved into the EXIF. Once every box is photographed and described, an AI agent assembles it all into a searchable index.

Prerequisites

Before you start, you need:

1. Photos enriched with KontexVoX. One photo per box, with the number and contents in the EXIF Description field. See What is an EXIF file?.
2. An AI agent with file-system access: Claude Code, Codex CLI, or Gemini. Not a web chatbot.
3. The photos exported to your computer via AirDrop or the Files app (the channels that preserve EXIF — see How to share without losing the EXIF).
The index needs each box's description. If the EXIF is erased by an upload to a web chatbot, the index is empty. You need an agent that reads the files locally.

What you dictated: an example

Here's what the EXIF of four box photos contains:

IMG_0021.jpg → "Box 14: coffee maker, filters, blue mugs, Italian cookbook, dish towels. Kitchen." IMG_0022.jpg → "Box 15: fitted sheets (queen), pillows, summer duvet. Bedroom." IMG_0023.jpg → "Box 16: Bosch drill, screwdrivers, box of wall plugs, level. Garage." IMG_0024.jpg → "Box 17: grandma's photo albums, grandpa's letter, wooden frame. Handle with care. Living room."

These descriptions, you dictated while looking at the open box, just before taping it shut. KontexVoX transcribed and engraved them into the file. Three months later, when you're looking for the coffee maker, the index finds "Box 14" in one word.

The prompt

Read all the JPEG images in this folder. For each image, extract the EXIF Description field (which contains the box number and its contents). Generate a self-contained HTML file with a search bar. When you type a word (e.g. "coffee maker"), the app filters the boxes whose description contains that item. Display the box number, the full list of its contents, and the photo thumbnail. The HTML must be responsive and work offline.

What you get

A single HTML file, openable on any device. Put it on your phone, type "coffee maker", the index says: Box 14. You go to box 14, open it, the coffee maker is there. No digging, no wasted time.

The file works offline — no connection, no account, no cloud. Once generated, it's self-contained. You can share it with your partner, your kids, the people helping you unpack. Everyone has the index on their phone.

Recommended tool: Codex CLI

Codex CLI is ideal for generating an index file, CSV or HTML, with a search function. The self-contained HTML format is practical: one file, openable on any device.

Frequently asked questions

Can I export the index as CSV instead of HTML?

Yes. Add to the prompt: "Also generate a CSV file with columns: box number, contents, room, photo file name." The CSV is useful for sorting in Excel or Google Sheets. The HTML is handier for searching on a phone.

How many boxes for a standard move?

Between 30 and 80 boxes for an apartment or a house. With KontexVoX, you take one photo per box as you close it — 50 enriched photos take 20 minutes. The AI agent generates the index in seconds. The documentation work, which would have taken an evening by hand with a marker, happens at packing time.

Does the index work if the boxes are in several places?

Yes. Add the place to your description: "Box 14: coffee maker, filters. Mother-in-law's garage." The index will show the place with the number. You'll know not only which box, but where the box is.

Can I do the same thing without KontexVoX?

Technically yes — you can type each box's contents into a text file, then run the prompt. But 50 boxes means 50 descriptions to type. Voice changes everything: you say the contents while looking at the box, without letting go of the tape. That's why I created KontexVoX — not for moving, but so the things we document while doing them don't fade away.