Generic titles, thin descriptions, and missing attributes cost you impressions and clicks. Trajekt's AI Rules use Claude AI to rewrite product data at scale — improving titles, enriching descriptions, extracting missing attributes, and generating channel-specific copy that outperforms your competitors.
✓ Claude AI powered✓ Per-product credit pricing✓ Full catalogue processing✓ Rules-based targeting
30%
average CTR improvement from AI-enriched product titles and descriptions
10,000
products enriched per hour with Trajekt's AI Rules engine
£0.001
typical cost per product for basic AI title and description enrichment
100%
of your AI enriched output is validated before being included in feeds
Features
AI Product Feed Rules
✍️
Title rewriting
AI Rules can rewrite product titles to include brand, material, colour, and size in the optimal order for each category — at scale, across your entire catalogue.
📝
Description enrichment
Thin descriptions kill Shopping performance. AI Rules generate factual, keyword-rich descriptions from your product title, category, and attributes — suitable for both Google Shopping and ChatGPT.
🎯
Attribute extraction
AI can extract colour, material, style, and other attributes from unstructured product titles and descriptions — populating missing feed fields automatically.
🧠
Channel-specific copy
Write different descriptions for different channels. A Google Shopping description is different from a ChatGPT-optimised one. AI Rules can generate both from the same source product.
💰
Cache & credits
AI results are cached per product — if a product doesn't change, the cached result is used and no credits are consumed. Credits are only used for new or changed products.
📊
Rules targeting
Target AI Rules at specific product types, categories, or conditions. Only rewrite descriptions for products with less than 50 words. Only extract colour for apparel. Full control.
FAQ
Frequently asked questions
AI product feed rules are instructions that apply Claude AI to specific fields in your product feed at scale. For example: "Rewrite the description for all products with fewer than 100 words in the description field, making it factual and keyword-rich for Google Shopping." Trajekt processes your entire catalogue through these rules automatically.
Trajekt uses Claude (by Anthropic) for AI feed rules. Claude is particularly well-suited for product data enrichment because of its ability to follow precise instructions, maintain consistent formatting, and generate factual content without hallucination.
Trajekt uses a credit system for AI Rules — typically fractions of a penny per product. A basic description rewrite costs approximately £0.001 per product. For a 10,000-product catalogue, a full description refresh costs approximately £10. Credits are purchased as needed and results are cached to avoid re-processing unchanged products.
When Trajekt processes a product through an AI Rule, the result is cached against a hash of the product data. If the product hasn't changed in your next feed run, the cached result is used and no credits are consumed. This means ongoing feed costs are very low — only new or updated products consume credits.
The most impactful AI rules for Google Shopping: 1) Title rewriting to include brand + key attributes in optimal order, 2) Description enrichment adding missing specifications and use cases, 3) Attribute extraction for colour, material, size from unstructured text, 4) Category improvement — better google_product_category assignment. These directly impact impression share and CTR.
Yes — and this is where AI enrichment adds the most unique value. ChatGPT evaluates product descriptions for factual completeness, clarity, and relevance to conversational shopping queries. AI Rules can rewrite descriptions specifically for conversational AI audiences — factual, structured, without marketing hyperbole — dramatically improving ChatGPT product matching quality.
Trajekt's rule targeting lets you apply AI Rules based on conditions: product type, category, description word count, brand, availability, or any other field. For example: apply description rewriting only to products where description contains fewer than 100 characters. This lets you focus AI credits on products that need enrichment most.
Standard rules are deterministic — if title contains "jacket", prepend brand name. AI rules are generative — use AI to rewrite the description to be more suitable for Shopping channels. Standard rules are instant and free. AI rules consume credits but can make nuanced improvements that pattern matching cannot. Trajekt supports both, and you can combine them.
Processing speed depends on catalogue size and rule complexity. Typically, 1,000 products can be processed in 5-10 minutes. For initial setup with a large catalogue, Trajekt processes in batches and caches results. Subsequent runs only process changed products, making them much faster.
This is a valid concern. Trajekt mitigates hallucination risk by: 1) Constraining AI Rules to only use data present in the source product fields, 2) Using structured prompts that instruct Claude to only use provided information, 3) Validating output before inclusion in feeds. AI Rules work best for enrichment tasks (improving existing data) rather than invention tasks (creating data from nothing).
Yes. You can write your own AI rule instructions in Trajekt. The rule editor lets you write the exact prompt that Claude follows, specify the source fields to use, set the output field, and define targeting conditions. This gives you full control over the AI enrichment output.
Connect your store and Trajekt's AI Rules engine improves titles, descriptions, and missing attributes across your entire catalogue — credits from 0.1p per product.