24 Hours of Art

PublishingAI PipelineStripe · Token Gating

24 Hours of Art had published a daily digital and crypto art sales report for more than 970 consecutive days, composed by hand inside an email client every morning. I rebuilt it as a database-backed publication with an AI enrichment pipeline sitting under the admin: it ingests the day's sales from Roger's spreadsheet, finds artwork images and buyer, seller, and collection links, converts crypto amounts to USD, and pre-populates every section of the report before he touches it. The public site added things the email could never do, including a collage image where the system detects individual artwork boundaries and turns each one into a hoverable sale record, an automated market signals ticker, and a subscriber tier that combines wallet-based token gating with Stripe.

VISIT THE SITE · beta.24hours.art ↗

970+Daily reports, no misses

10+Automations

2Paid access paths

Problem

Roger at 24 Hours of Art has published a daily sales report on digital and crypto art for over 970 days without missing one. He wakes at 4 a.m., researches the previous day's sales, and sends by 8 or 9 a.m. The entire report was typed directly into an email program. There was no database behind it, no archive, no way to make the data interactive, and no structure to build a paid product on top of.

Approach

The rule was that AI removes assembly work, not editorial judgment. Every field the pipeline populates lands in the admin as an editable draft, so Roger keeps final say on the record while skipping many of the manual processes and data collection that ate his morning. That distinction shaped the build. The system pre-populates aggressively and overrides nothing.

What I shipped

An admin and a public site. The admin pulls the day's sales from Roger's spreadsheet, then an AI pipeline enriches each record: artwork images, links to the sale platform, buyer and seller links, collection data, and USD conversion from the ETH price.

Market signals run automatically across roughly a dozen collections Roger flagged as market indicators, pulling one-day and year-to-date price and percentage changes plus imagery. He picks which ones publish and which one is featured, and the featured collection pulls thumbnails of its latest sales. A ticker across the homepage carries all twelve.

The image grid is the piece I like most. Roger uploads one collage of the day's artwork. The system finds the boundaries of each individual work inside that single image and draws editable boxes, which he maps to the corresponding sale or annotates himself. On the homepage, hovering a region highlights the sale, surfaces amount, buyer, and seller, and links out to the platform listing.

Access is tiered. Wallet connect with token gating for holders of 24 Hours tokens, Stripe for paid subscriptions. Free and signed-out visitors get the grid, top sales, and headlines. Pro subscribers get the full day of sales, split by category into a filterable, sortable list with imagery, USD amounts, and buyer, seller, and collection links. Subscribers also get monthly roundups, which auto-compile the month's top sales and headlines and include interviews, takeaways, and highlights.

Outcome

A daily email typed into a mail client is now a structured, queryable publication with an archive, a sponsor placement, and two paid access paths. Roger's morning shifted from copy/pasting art sales into an email to working within a streamlined, AI-assisted pipeline that produces clean, enriched archive data.

← ALL WORK START A PROJECT →