For appraisal professionals: insurers, court experts, architects, valuation firms. You're in the field, photographing a construction site, a loss, or a property. Every photo is described out loud: what you see, the measurements, the defects. KontexVoX engraves the description into the EXIF, with date, time and geolocation. Now you need to turn this set of photos into a professionally formatted appraisal report.

An appraisal report is a formal document: summary, detailed observations, conclusion. Every observation is documented with its reference photo, timestamp and geolocation. The report must be factual, neutral, defensible. Writing it by hand from 50 photos is a day's work. With the descriptions already in the EXIF, the AI agent assembles the report in minutes.

Prerequisites

Before you start, you need:

1. Photos enriched with KontexVoX. The descriptions live in each file's EXIF Description field, with capture date and geolocation. 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).
An appraisal report is rigor: numbered observations, timestamps, geolocation, factual tone. A web chatbot doesn't structure that kind of document from 50 photos. You need an agent that reads the batch and holds the professional tone.

What you dictated: an example

Here's what the EXIF of four appraisal photos from a construction site contains:

IMG_0101.jpg → "Foundation, northwest corner. Vertical crack, 2mm, 80cm tall. Visible water penetration." — 06/18/2026 14:22 — GPS: 48.3567, 2.4012 IMG_0102.jpg → "Roof frame, central truss. Joist cracked at the joint. Damp wood, traces of fungus." — 06/18/2026 14:31 IMG_0103.jpg → "South facade. Render flaking over 4 m², exposed to runoff." — 06/18/2026 14:38 IMG_0104.jpg → "Roof, east slope. Three tiles missing, underlay visible. Infiltration stain on the battens." — 06/18/2026 14:45

These descriptions, you dictated while walking the site, point by point. KontexVoX transcribed, timestamped and geotagged them. The AI agent will structure the observations, number them, and produce the report.

The prompt

Read all the JPEG images in this folder. For each image, extract: the EXIF Description field, the capture date and time, the GPS geolocation. Generate an appraisal report in Markdown structured as follows: 1) Header with date and location of the appraisal, 2) Summary of observations, 3) Detail per observation (numbered, with description, timestamp, and geolocation), 4) Conclusion. The tone must be professional, factual, neutral. Include an appendix with the complete list of photos and their metadata.

What the AI produces with this prompt

An appraisal report in Markdown, structured in four sections:

1. Header: date and location of the appraisal, extracted from the metadata.
2. Summary: the main observations in a few lines.
3. Detail per observation: each observation is numbered, with its description (your dictated text), its timestamp and geolocation. Observations are grouped by area (foundations, roof frame, facade, roof).
4. Conclusion: the AI does not draw conclusions for you. It leaves an empty section for your expert opinion.

The report converts to PDF. The tone is factual and neutral, as expected in an appraisal document. The reference photos with their EXIF metadata intact remain the documentary evidence.

Recommended tool: Claude Code

Claude Code is ideal for generating long, structured, professional documents. For an appraisal report that must be rigorous and well organized, it's the best choice.

Frequently asked questions

Does the report replace my expertise?

No. The report documents the observations you dictated. The summary and conclusion are yours to write — the AI leaves the conclusion empty for your expert opinion. The report saves you the layout and structuring time, not the analysis time.

Can I customize the structure?

Yes. If your firm expects a specific format (mandatory sections, different numbering), describe it in the prompt: "Structure: 1) Property identification, 2) Methodology, 3) Observations by area, 4) Diagnosis, 5) Conclusion." The AI follows your structure.

Is the report defensible?

The report is a structured, traceable working document. It has no legal force in itself, but it documents the chain of observations with timestamps and geolocation. The original photos with intact EXIF metadata remain the evidence. In a dispute, the report is support material, not a substitute for a formal procedure.

Can I do the same thing without KontexVoX?

Yes — you can take photos and write the report by hand. But in the field, your hands are busy: camera, tape measure, flashlight. Voice lets you describe without putting the camera down. That's why I created KontexVoX: so the observation happens at the moment you see the defect, not that evening at the office.