Live, limited P1

Train store voice and image rules

Analyze material you own, review extracted suggestions, and save only useful store-scoped rules.

Availability
Live, limited
Last verified
Runtime anchor
ab084f79c5ed3fc6a28f07c64b29920944a0e5db
Certification anchor
db23559179dc3c67be8a1ab16a0f07323b9820f7
Evidence note
Source admission, safe fetch, Reddit export, OCR, and store-scoped save contracts were verified; analysis quality remains input-dependent.

What this does

Analyze material you own, review extracted suggestions, and save only useful store-scoped rules.

Before you begin

  • Select the exact store before analysis; saved rules follow that store.
  • Use only writing, sites, exports, and images you have the right to process.

What changes outside ArtDrop

Voice Trainer and Image Trainer can upload content, safely fetch one website page, run OCR, and call analysis services. Saving changes store-scoped copy rules; it does not publish a product.

ArtDrop Brand Voice workspace on the Upload writing trainer tab
Train from owned writing. Upload writing you are allowed to process, inspect the extracted rules, and save only the guidance that belongs to the selected store.
ArtDrop Brand Voice workspace on the Website trainer tab with detected-store assistance
Train from a website. Analyze an authorized public site as a voice source, then review every inferred rule before it becomes store-scoped guidance.
ArtDrop Brand Voice workspace on the Store identity trainer tab
Describe the store identity. Store identity fields provide explicit context for the selected store; they do not copy credentials or silently affect another destination.
ArtDrop Brand Voice image trainer for refining generated artwork copy through feedback
Image trainer. Analyze one owned artwork, refine the draft in plain language, and save only reusable rules for the selected store.

Steps

  1. Choose a source

    Use Upload Writing for DOCX, PDF, TXT, CSV, JPG, JPEG, or PNG; Website for an About/bio/statement page; or Store Identity for direct context.

  2. Use Reddit export correctly

    Upload your own Reddit GDPR comments/posts CSV. Public username scraping is not supported.

  3. Review extraction

    Inspect voice, content, example, and never-use suggestions. OCR and website analysis can be incomplete or wrong; edit them before saving.

  4. Train image interpretation

    In 2 Image Trainer, analyze one owned artwork, refine the result in plain language, and save only reusable rules to the selected store.

Expected result

The selected store shows reviewed voice and image rules, and a later draft reflects them without copying credentials across stores.

Limits and non-goals

  • Up to 10 source files, 25 MiB each and 32 MiB combined; PDFs up to 50 pages. OCR has stricter practical limits.
  • Website fetches are bounded and do not follow arbitrary redirects; some sites cannot be analyzed.

Troubleshooting

  • If OCR misses stylized text, supply an accessible text version instead of repeatedly enlarging the image.
  • If a website fails safe-fetch checks, upload the owned text as a file.

How to undo or clean up

Edit or replace saved rules in the same store. Copying settings to another store copies voice and lines only when you explicitly choose that action.