Skip to content

Type a term, e.g. API, Shopware or SEO.

    ← All projects

    10 · AI automation · SEO · GEO · E-commerce

    10,000 items — new copy, new images.

    A reseller with around 10,000 items from dozens of brands barely ranked in search engines — because nearly all product copy and images had been taken over directly from the manufacturers.

    Client project · client anonymized (confidentiality) · project from 2026

    • AI text generation
    • AI image generation
    • Nano Banana Pro
    • SEO
    • GEO
    • Custom development
    • PHP middleware
    • MariaDB
    10,000
    items with completely new copy and images
    0%
    duplicate content once the pipelines finished
    own check using a language model
    2 pipelines
    text + image — generated automatically, reviewed before going live

    01 · Problem

    The client runs an online store as a reseller for around 50 well-known brands and carries approx. 10,000 items. As is common among resellers, product copy, short descriptions and images were taken over directly from the manufacturers — identical to thousands of other stores carrying the same brands. From an SEO perspective, a major problem: duplicate content on nearly every product page, hardly any relevance of its own for search engines, weak rankings despite a decent product range. On top of that, the store was not geared at all to the requirements of AI-powered search systems such as ChatGPT, Perplexity or Google AI Overviews — it was barely visible in AI search.

    02 · Solution

    vona built two fully automatic AI pipelines that rebuild the store's entire catalog independently of each other — and around the clock.

    Pipeline 1 — copy: A PHP middleware reads each item from the store database, passes the product name, category and original manufacturer text to the AI and uses them to generate fully original, SEO-optimized product copy, short descriptions, meta descriptions and page titles — based on defined prompts that already incorporate tone of voice, keyword strategy and GEO requirements. All copy is written straight back into the store database.

    Pipeline 2 — images: A second pipeline uses Nano Banana Pro to generate new, original product images based on the original product images — visually consistent, high-quality and no longer identical to the manufacturer assets. For every newly generated image, the pipeline automatically creates SEO-optimized alt text and assigns both to the respective item.

    Review before publishing: copy and images do not go live unchecked. Every draft is reviewed before publishing – on the same principle we use to check the entries in our AI tools directory before they appear.

    03 · Result

    Both pipelines generate the new copy and images for the entire catalog of approx. 10,000 products automatically; each item is published after review. Once both runs were complete, our check using a language model found no item left with manufacturer-supplied copy or images. For the first time, the store has a content identity of its own — every product page is unique, relevant to search engines and geared to AI-powered search systems. Newly listed items pass through both pipelines and the same review — so they get their own copy and images too.

    04 · Workflow

    1. Store database: approx. 10,000 items from around 50 brands
    2. PIPELINE 1 · COPY: Middleware reads (name, category, manufacturer text), AI rewrites (prompts with tone, keywords and GEO), Original copy (product text, description, meta description, title)
    3. PIPELINE 2 · IMAGES: Original images (from the manufacturers), Nano Banana Pro (new, original images), Image + alt text (SEO-optimized, assigned to the item)
    4. Unique items: no manufacturer copy, no manufacturer images
    5. NEW ITEMS PASS THROUGH BOTH PIPELINES AUTOMATICALLY
    Two pipelines rebuild the entire catalog of around 10,000 items, independently of each other and around the clock: one writes original copy, the other generates new images with alt text. The result is written back to the store database. Newly listed items pass through both pipelines automatically, which rules out duplicate content for good.

    Next project

    Your project here?

    Tell us about your problem. The first conversation is of course free of charge and without obligation.

    Request a project →

    Newsletter

    News from the vona workshop.

    AI tools we actually use, our weekly AI review & insights from our projects – short, practical and without spam.