Content Automation: AI-Generated Descriptions for 1C-Bitrix

Content Automation: AI-Generated Descriptions for 1C-Bitrix

Every online store faces the same challenge: hundreds or thousands of products without unique descriptions. Copying competitors' texts risks penalties; manual writing is slow and expensive. The solution is integrating an AI model with Bitrix infoblock. We've set up this pipeline for 50+ projects: the system pulls product characteristics from the infoblock, builds a prompt, and returns the finished text to the product card. The time to write one description drops from 20 minutes to 20 seconds, and the content budget is reduced by up to 80%.

Our team consists of certified engineers with over 5 years of hands-on experience. We've automated catalogs ranging from household appliances to industrial equipment. REST API and infoblocks v2.0 are our daily stack. The result: organic traffic to product cards grows up to 34% within 3 months after indexing. Contact us — we'll demonstrate how it works on your data. Order an audit of your current state, and we'll propose an implementation plan.

How We Collect Data for the Prompt

The data collection process consists of several steps:

1. Retrieve the infoblock element via CIBlockElement::GetByID.
2. Extract fields (name, section) and all properties, excluding technical ones like MORE_PHOTO.
3. Add the current price via CCatalogProduct::GetOptimalPrice.
4. Build a context string with line breaks.

The quality of the text is directly proportional to the quality of input data. Example implementation:

function buildProductContext(int $elementId): string { $element = CIBlockElement::GetByID($elementId)->GetNextElement(); $fields = $element->GetFields(); $props = $element->GetProperties(); $context = "Product: {$fields['NAME']}\n"; $context .= "Catalog section: " . getSectionName($fields['IBLOCK_SECTION_ID']) . "\n"; foreach ($props as $prop) { if (!empty($prop['VALUE']) && $prop['CODE'] !== 'MORE_PHOTO') { $context .= "{$prop['NAME']}: {$prop['~VALUE']}\n"; } } // Add price for context $price = CCatalogProduct::GetOptimalPrice($elementId); $context .= "Price: {$price['PRICE']['PRICE']} {$price['PRICE']['CURRENCY']}\n"; return $context; } 

Properties like MORE_PHOTO and other technical ones are skipped — only semantically significant data should be included in the prompt. As the 1C-Bitrix documentation states: "An infoblock is a universal way to store structured information."

Multi-Level Prompt System

Not one prompt for all products — separate templates for each product category. Electronics and children's clothing require different tone and structure.

The prompt system with inheritance:

• Base prompt: general instructions on style, forbidden phrases, and structure.
• Category prompt: specifics of the category (technical language for electronics, emotional for lifestyle products).
• Infoblock prompt: store specifics (brand's unique voice).

If a category prompt is not set, the prompt from the parent category is used, then the base one.

Formatting and Output Structure

Asking AI to return pure HTML text is unreliable — the model may break the structure. It's better to request structured JSON:

{ "preview_text": "Short description up to 100 words for listing", "detail_text": "Full description 200–350 words with HTML formatting", "bullet_points": ["key advantage 1", "key advantage 2"], "target_audience": "Who this product is for" } 

JSON mode is available in OpenAI via response_format: {type: "json_object"}. Bullet points are used for the "Advantages" block on the product card via an infoblock property of type S with the MULTIPLE flag.

How to Organize Batch Generation?

GPT-4o-mini has a 128K token context window. You can send a batch of several products in one request — this reduces overhead from system tokens:

Describe the following 5 products. Return a JSON array of 5 objects... [product 1] [product 2] ... 

A batch of 5 products saves up to 30% of tokens on system instructions. But if an error occurs, the entire batch is lost — we implement retry with escalation and limit batches to 3–5 products maximum.

Why Is Prompt Testing Important?

For highly competitive categories, it's worth testing different prompts:

• Group A: technical style (specifications → benefits)
• Group B: emotional style (lifestyle → technical details)

Bitrix supports A/B testing via the marketing module, but it's easier to store the prompt variant in a product property and measure conversion through goals in Yandex.Metrica.

Case Study: Generating Descriptions for 28,000 SKUs from Our Practice

Challenge: household appliances, 3 categories (large, small, climate), different text requirements. Implementation:

• 3 category prompts developed with a marketer in 2 days
• GPT-4o-mini for the bulk (80%), GPT-4o for the top 100 SKUs by revenue
• Batches of 3 products, 8 parallel workers
• Total generation time for 28,000 descriptions: 14 hours

Result: organic traffic to product cards increased by 34% within 3 months after indexing. Content budget savings up to 80% compared to manual writing. Request a consultation — we'll select the optimal configuration for your store.

What's Included in the Work

• Prompt system design: analysis of your catalog, development of base and category templates
• Infoblock integration: context collector, data transmission to AI, writing results back to element properties
• Batch generator: queue task management, retry processing, error logging
• Quality control: moderation of the first 100 descriptions, checklist setup
• Training and documentation: system handover, operational instructions, one-month support

Timeline

Total: 5–9 business days until first productive generation.

Get a consultation on content automation for your store — our engineers will help you choose the optimal solution. Contact us, and we'll show you how it works on your data.