Arbor & Eye

Independent design study, uncommissionedSeptember 6, 20262 min read

I responded to a public brief about an artist’s Shopify store by building a separate gallery for twenty-eight watercolors. Each work has its own view and enquiry path, so the artist could inspect the proposed experience.

The prototype includes all twenty-eight works, artist and process chapters, and enquiries without payment. I built it uncommissioned from a public Upwork brief. No client engagement followed.

Start with the decision

An independent, uncommissioned study; not the artist’s site and not a client engagement.

Full scope and limits

Uncommissioned prototype, built in July 2026 and refined that September. No client engagement followed.

Who did what

Me: I interpreted the brief, directed the design and delivered the prototype.
AI: Development and the empty room settings; the paintings are the supplied scans, placed without generative alteration.

The Arbor & Eye prototype presents Pulse of Spring in a dusk room, with a direct link to the work. — Open full screenshot

A decision and its outcome.

Situation: The public brief asked for refinement of an existing Shopify store: typography, spacing, imagery and a more personal, gallery-oriented feeling.
Decision: I chose to build a complete, separate prototype rather than propose changes to the live store.
Reason: I wanted the prospective client to be able to try the changes, instead of having to imagine them from a list. I also kept the study out of search indexes to avoid competing with the real site.
What changed: I delivered a complete gallery that could be compared with the store. No engagement followed, and I do not know whether the artist looked at it.

Responding to the public brief

The public request was for refinement of an existing Shopify website. It emphasized typography, spacing, imagery, and an experience connected to the artist. Tappan’s restraint was a reference, with a more personal, gallery-oriented feeling as the aim.

The supplied artwork and presentation material gave me enough to build the separate study. The following sections show how I used them.

Presenting the artist and the paintings

Ned Church is the subject of the site; Arbor & Eye is the studio and family mark. The biography and process keep his first-person voice. His father’s words retain their attribution, and the Tree of Life insignia has its own chapter after the work and the artist.

I directed a salon-wall presentation for the twenty-eight watercolors. Each opens a view with its own address, artwork details, and previous/next navigation. The catalog reproductions stay at modest sizes, respecting the resolution of the supplied images.

The first two rows of Arbor & Eye’s salon wall, with individually selectable watercolors arranged by shape and color. — A detail of the working gallery. Each painting opens its own view and enquiry path. Open full image

Using the supplied scans

I used the supplied scans for the paintings. AI generated the empty room settings, and the artwork was placed into them locally without generative alteration.

Only the known works carry supplied names and specifications. The remaining paintings use catalog numbers and details on request. I kept prices, editions, availability, and unresolved delivery times for the studio to confirm.

Enquiring about a work

An original work has a direct enquiry link. A visitor interested in prints can gather works in a bag, choose a destination, and prepare one email with the details. The bag opens over the gallery so the paintings remain close at hand; checkout puts the selected work and form first.

The prototype collects no payment. The bag stays on the visitor’s device, and the account preview transmits no email address. The studio would still need to confirm specifications, price, availability, and delivery.

The compact Arbor & Eye checkout with one selected watercolor, a shipping destination, and labeled delivery fields. — A print request, with confirmation still in the studio’s hands. No payment is collected. Open full image

Return to all work

Next project

Receiver Reliance

I studied when an AI agent should rely on another’s work. Across 1,805 synthetic episodes, a structural check cut improper handoffs by an estimated 1.0 per episode against transparent release, with an interval from −1.6 to −0.4. It also completed fewer seeded tasks. Wrong answers with coherent paperwork still passed.