Case study: from R scripts to an art valuation platform

Client

Art advisory & valuation firm (anonymous)

Project type

Internal data & valuation web platform

Status

In production since 2025

The challenge

The firm valued artworks with its own model written in R and spreadsheets filled by hand with auction results. Capturing every lot, converting currencies, finding comparables and building the charts was manual, repetitive work.

What we built

• ✦Copy-paste capture: the team pastes an auction record and the system extracts artist, title, medium, dimensions, hammer and premium prices, estimates, auction house and date, converting EUR and GBP to USD.
• ✦Comparable-sales valuation: price per cm² for two-dimensional work and per cm³ for sculpture, with configurable year window, commission and currency.
• ✦Same model, no surprises: the engine replicates the logic of the original R scripts, so new results are comparable with historical ones.
• ✦Comparative analysis: CAGR and total return between sales, 10-year price behavior and supply versus demand.
• ✦Duplicate detection so the same work isn't counted twice, even when its title is written differently.
• ✦Automatic backups of the database and images to Google Drive every 4 days.

Decisions that matter

• ✦Not every problem needs a chatbot. The value was in automating capture and standardizing the model, not in a conversational interface.
• ✦Respect the client's know-how. We replicated their model before improving anything, so the team trusted the numbers from day one.
• ✦Security from the start: hashed passwords, login attempt limits and rate limits on every endpoint.

What it means for a sales team

Quotes, proposals and reports often live in spreadsheets someone fills in by hand. The principle is the same: capture the data once, calculate with clear rules and let the team spend its time selling. To find that repetitive work in your operation, take the free AI Assessment.