AI for Printmaking Edition Records

As of 2026-10-19 · the AZET team. We operate azet.io and write this educational guide for fun and reference. The print session, sheet labels, and ledger discrepancies below are hypothetical examples, not documented studio results.

A sheet can have a permanent studio ID before it has an edition label. Keeping those two things separate makes a small print session easier to reconcile. You can record what exists, what the maker has designated, and what remains undecided without filling the blanks by guesswork.

This guide proposes a local record system, not a rulebook for edition terminology. The maker supplies the meaning of any labels used. An AI text assistant's bounded task is to compare those supplied records and list inconsistencies for a person to resolve beside the physical sheets.

Give every sheet an identity independent of its status

Assign a stable sheet ID, then record the session, image version, maker-assigned label, observed difference, and storage location. Put any planned edition size in a separate session field. A plan describes an intention; it should not create records for sheets you have not accounted for.

Use “undecided” for a status the maker has not assigned. That differs from a blank caused by missing transcription. If a handwritten note is unreadable, say so and retain a reference to the note. The assistant should not transform unclear handwriting into an edition decision.

For this exercise, the maker defines three local statuses: included, study, and undecided. These are simply the categories chosen for the invented ledger. No claim is made that other studios should use the same vocabulary.

A worked reconciliation with a duplicate label

Imagine a hypothetical session called Window Block, image version V1. The maker's plan lists four included sheets. A physical inventory accounts for six sheets, with the following supplied records:

Sheet ID	Local status	Recorded label	Location
W01	Study	None assigned	Folder A
W02	Included	1/4	Folder B
W03	Included	2/4	Folder B
W04	Included	2/4	Folder B
W05	Included	4/4	Folder B
W06	Undecided	None assigned	Folder C

The useful result is a short exception list: the label 2/4 appears twice, 3/4 is absent from the recorded labels, and W06 has no maker-assigned final status. The physical inventory contains six sheets. None should disappear to make the ledger look like the four-sheet plan.

Do not automatically relabel W04 as 3/4. The pattern suggests something to inspect, not an authorized correction. The mistake could be in transcription, on a label, or in the maker's earlier decision. Those possibilities require different records.

For an illustrative resolution, suppose the maker checks the physical W04 sheet and its original session note, then supplies: “Both read 3/4; the ledger was transcribed incorrectly.” The corrected ledger can now show 3/4 while retaining a correction entry stating the old value, new value, basis, and person who checked it. W06 remains undecided.

Ask for exceptions before a clean summary

Reconcile this print-session ledger with the supplied physical
inventory and the maker's stated local definitions.
Keep sheet IDs separate from edition labels and planned totals.
List duplicate labels, missing values, conflicting image versions,
and mismatches between the inventory and ledger.
Do not assign edition numbers or decide which sheets are included.
Separate confirmed records from unresolved discrepancies.
For each discrepancy, name the physical sheet or source note a
human should inspect. Draft corrections only after I provide
verified replacement values, preserving the earlier value.
Inputs: [paste definitions, session plan, inventory, and ledger].

Requesting exceptions first keeps the assistant from smoothing over the interesting part. A clean table can be produced after discrepancies are resolved or explicitly retained. Its neatness should come from reliable entries, not silent assumptions.

Keep visual variation separate from administrative status

Suppose the hypothetical note for W03 says “pale patch near lower left.” Record that phrase in the observation field. It does not automatically change W03 from included to study. The maker decides status using their own criteria and records that decision separately.

Likewise, a revised image version needs its own explicit field. If a later sheet uses V2, do not let a shared session name hide the difference. Ask the maker how that sheet belongs in the record rather than treating all sheets with the same title as interchangeable.

For a final human check, count the physical sheets by stable ID, match locations, and inspect the assigned labels. Recalculate category totals from those checked rows. Keep planned totals visible beside actual records so an incomplete plan cannot masquerade as a completed edition.

FAQ

Should study sheets be deleted from the ledger?

No, not in this proposed system. They remain physical sheets with identities and locations. A filtered view can show only included sheets while the complete ledger preserves the session inventory and the maker's choices.

Can AI choose the missing edition number?

It can identify a gap in supplied labels, but this exercise reserves assignment for the maker. A missing number does not establish which sheet should receive it or whether an earlier entry is wrong. Inspect the object and source notes first.

How should I record a later correction?

Retain the earlier value and add the verified replacement with its reason and date. Distinguish a transcription correction from a new maker decision. The current table may show the latest value, while the correction note explains how it got there.

Product baseline

This workflow is a general illustrative exercise, not a documented AZET capability. Our maker introduction describes an AI assistant and an email waitlist as of October 3, 2026. It provides a historical product baseline, not confirmation of current availability or print-record management features.

For the AZET waitlist, visit azet.io.