Skip to content
P2Lab AI

Generate content

Generate Content writes name, description and meta fields from scratch for products or categories. Unlike Improve (which rewrites existing text), this feature is for filling gaps — products that arrived with no description, categories that need landing copy.

Three source modes

Section titled “Three source modes”

You pick the mode in the modal:

Model knowledge only

Section titled “Model knowledge only”

Pure LLM generation. The model writes only from the product attributes it can see (name, properties, categories, manufacturer). Fastest and cheapest, but limited to what the model already knows.

Section titled “Web search”

The model uses OpenAI’s built-in web search tool to look up real context — brand information, technical specifications, comparable products — before writing. This adds a separate per-call web-search charge, billed by the provider on top of the token cost, to every product.

A Search depth dropdown (low, medium, high) controls how many search results the model is allowed to consume.

URL fetch (custom field)

Section titled “URL fetch (custom field)”

You provide one or more URLs (typically read from a product custom field — manufacturer pages, datasheets, etc.). The plugin fetches each URL, extracts the relevant content using the matching fetch domain config, and feeds it to the model as context. Use this when you have authoritative sources you trust more than the model’s prior knowledge.

Target fields

Section titled “Target fields”

The modal lets you choose which fields to generate. For each field you can override min / max length, format specification, custom instructions, and reference products / categories:

  • name — 5–80 characters
  • description — 800–1500 characters recommended
  • metaTitle, metaDescription, keywords — same constraints as Generate SEO

The model returns HTML for description (no markdown). The plugin’s HTML normaliser cleans it up before applying.

How to use

Section titled “How to use”

The Generate content modal is typically launched from a bulk job because it generates several fields at once across multiple languages.

  1. Go to AI Assistant → Jobs → New job.
  2. Pick Generate content as the type and choose product or category.
  3. Step through the wizard:
    • Step 1 — Choose type.
    • Step 2 — Filter the records (categories, manufacturers, “missing description only”, etc.).
    • Step 3 — Configure:
      • AI connection
      • Source dropdown — Model knowledge only / Web search / URL fetch (custom field)
      • Search depth (only for Web search): low / medium / high
      • Custom field mapping (only for URL fetch): pick which product custom field holds the source URLs
      • Target languages (multi-select)
      • Fields to generate — for each field, optional min/max length, format spec, custom instructions, reference products/categories
    • Step 4 — Review the cost estimate and start the job.
  4. The wizard processes each record and produces one suggestion per (field, language). Review them on the jobs detail page and apply individually or in bulk.

Configuration

Section titled “Configuration”
  • Length constraints — override the defaults per field if you want shorter or longer copy.
  • Custom instructions — house tone, brand voice, banned words.
  • Reference products / categories — extra context the model can read alongside the current product.

Tips & gotchas

Section titled “Tips & gotchas”
Section titled “Related”