AI-Enhanced Product Enrichment in Akeneo PIM

Akeneo’s AI-Enhanced Enrichment feature transforms the way you generate, translate, and optimize product data. Using state-of-the-art generative AI, you can automate the generation and adaptation of product content from your attribute values and from your product assets (such as images and PDF), directly from the Product Edit Form or via automation rules.

What Can You Do with AI-Enhanced Enrichment?

• Generate new product descriptions using either product data (attribute values) or product assets (images, PDF).
• Translate product content across locales, with tone and brand consistency.
• Automate enrichment at scale through rules and targeted automation.
• Adapt existing content (rephrase, make more technical, adjust length, optimize for channels, etc.).

With the introduction of asset-based generation, you now have two main ways to create content with AI:

1. Generation from Attribute Values: Use existing product attribute data (text, selections, numbers, etc.) to generate or translate content.
2. Generation from Assets: Use product images and PDF (plus optionally, attribute data) to generate content based on the visual or document assets linked to your product.

Example Use Cases

• Attribute-based: Generate a product description for a t-shirt using its color, material, and size attributes.
• Asset-based: Generate a description for a lamp by analyzing product images (to detect design, color, shape) and a PDF datasheet (for technical specs).
• Multi-channel: Translate a product description from French to English while keeping the same tone and brand voice, optionally using images for context.

How to Activate AI Enrichment

Enable AI Role Permissions

Go to System / Roles / Permissions / AI and enable permissions for:

• AI on attributes
• AI usage in the Product Edit Form
• AI in Rule Engine

Enable AI at the Attribute Level

For each attribute you want to enrich, go to its settings in Settings > Attributes > Generative AI tab and choose which AI-powered actions to enable:

Attribute-based Generation

Enable This attribute can be enriched through AI to allow generation or translation from attribute data. Select which input attributes are used as context.

Asset-based Generation

Enable This attribute can be enriched from an asset to enable AI to use linked images and/or PDF (and optionally, some attribute values) as context. Choose the asset collections to be used and, if needed, select additional data fields.

Enable AI at the Family Level

Add specific prompts or tone/voice instructions for different product families for a more tailored result.

How to Use AI Enrichment

In the Product Edit Form (PEF)

1. Open a product and find an attribute enabled for AI enrichment.
2. Click the Ask AI button.
3. Choose your generation mode:
   • From Attributes: The AI uses your configured attribute values to generate content (e.g., title, description, select value).
   • From Assets: The AI uses selected product images and/or PDF (and optionally attribute data) to generate content. You can select the asset collections to be used.
4. Review the generated result and use available quick actions to refine it (shorter/longer, more descriptive, etc.).

In the Rules Engine

1. Go to Settings > Rules > Create a new rule.
2. Define product selection (filters/conditions).
3. Add an action:
   • Generate attribute value with AI: Generate from attributes only.
   • Generate from asset: Generate using assets (images/PDF) and optionally attribute values.
   • Generate translation with AI: Translate attribute value to one or several locales.
4. Configure the target attribute, channel, and locale.
5. Enable your rule. It will be executed automatically at scheduled times.

Limits

• Supported asset types: PDF, JPEG, JPG, PNG, WEBP, TIFF, and GIF (single frame).
• Supported number of assets per product: Up to 5 images (20MB each max) and 1 PDF (max 50MB, first 20 pages) per product.
• Asset collections: Only those with a “main media” attribute are supported. Kindly note that the “media link” is not yet supported; please bear with us!
• Processing order: The first images/PDFs in the selected asset collections are prioritized.
• Number of attribute options supported: 500. When selecting simple select or multi-select attribute types as output, please note that only up to 500 attribute options are supported due to prompt size limitations imposed by OpenAI. If the number of options exceeds 500, any additional options beyond the 500th will simply be skipped.
• Fair usage: Akeneo monitors usage and billing. Current limits are financial (per our AI provider plan), not quantitative; however, we will introduce limits per customer in the near future.
• Security: All data is processed through OpenAI’s Enterprise plan, with no training or storage of your data.

Best Practices & Troubleshooting

• Use high-quality, relevant assets for asset-based generation to ensure the AI produces accurate and detailed content.
• Refine your prompts to specify what the AI should focus on (e.g., “Describe the main features visible in the images, emphasize the material and shape”).
• Always review AI-generated content before publishing, especially for compliance and accuracy.
• If a field or option is unavailable, ensure the right permissions and feature flags are enabled and that you’re using supported asset/attribute types.
• For any errors or unexpected results, try reducing the number/size of assets or simplifying your prompt.

Security & Data Privacy

Akeneo takes your data security and privacy seriously. All product enrichment—whether generated from attributes, from assets (images, PDFs), or translations—is governed by robust security standards and privacy controls. We use OpenAI Enterprise and Google Gemini Enterprise, ensuring that the data we provide isn't utilized in model training. This data flows exclusively within their servers, returning directly to us without storage or reuse. We share only precise information with OpenAI to ensure that we generate accurate content tailored to your requirements.

AI models used

We are currently using OpenAI's ChatGPT models:

• gpt 4.1-mini: Text content generation — produces product descriptions, marketing copy, etc.
• gpt 4.1-mini: Text translation — translates text from one locale to another
• gpt 4.1-mini: Text rephrasing — rewrites existing text to improve style, tone, or consistency
• gpt 4.1-mini: Image analysis — processes images to extract product information, detect patterns and/or verify attribute consistency;

Information shared with our AI provider

For content generation

• Product name: the value of the attribute used as label
• Locale: the language of the generated content
• Values: the values of the product in the selected attributes to take into account
• Prompt: whether it is a pre-defined Akeneo prompt or your custom one that you have defined 

For content translation

• Source locale
• Target locale
• Value to translate
• Prompt: your custom one that you have defined 

For content rephrasing

• Value to rephrase
• Rephrase type 

Additional Security Measures

• Data is transmitted over secure, encrypted channels.
• No personal data or credentials are ever sent to the AI service unless explicitly included by the user.
• You have full control over which attributes, assets, and prompts are used and shared.
• Feature flags and permissions allow granular control of access and usage per user and per customer.

Compliance & Recommendations

• Always review AI-generated content before publishing for accuracy, compliance with company policy, and regulatory standards. You can use Workflows to do so.
• Do not include confidential or sensitive information in prompts or attribute values unless necessary and reviewed.
• If required, consult your legal or data protection team to ensure usage is in line with your organization’s privacy and data governance policies.

Prompt Guidelines

Creating effective prompts for generating product content with a model like ChatGPT requires clear instructions and an understanding of the desired tone, voice, and details you wish to convey. Here are guidelines to help you craft prompts that yield precise and engaging product content:

The more details you provide, the more the description will meet your expectations and the more it will be qualitative and distinctive. 

1. Identify key features

• List out the key features of the content that you would like to generate. These could include specifications, benefits, unique selling points, and any other relevant details that set it apart from similar items.

2. Specify the target audience

• Consider who your target audience is and what they care about. Tailor your prompts to resonate with their needs, interests, and preferences. 

3. Choose the tone and voice

• Define the tone and voice for the description. Tone could range from formal to informal, enthusiastic to neutral, or anywhere in between. Voice could be professional, friendly, whimsical, etc. Be specific about the tone and voice to ensure consistency in the product description.

4. Provide context

• Give context to the prompts by explaining the purpose of the content and how it will be used. This helps ensure that the prompts align with the overall goals and objectives of the project. 

5. Include any necessary technical specifications

• Provide relevant technical specifications. If applicable, include important specs that should be mentioned in the description (e.g., dimensions, material, compatibility). This is particularly important for tech products, appliances, and gadgets.

6. Mention any constraints or special instructions

• Outline any specific constraints or instructions. If certain phrases, keywords, or compliance requirements need to be included in the description, make them clear in the prompt.

7. Request examples or comparisons

• Ask for examples or comparisons if helpful. Indicating a request for the product to be compared to familiar items or scenarios can make the description more relatable and understandable to the target audience.

8. Clarify the purpose of the content

• Explain what the content generated will be used for. Whether it's for an online store, a catalog, or marketing material, knowing the context can help shape the content appropriately.

9. Review and refine the prompt

• Proofread your prompt for clarity and completeness. Before submitting your prompt, please ensure it clearly communicates all necessary information and instructions. You can test it on several products in the Product Edit Form before automating it in the Rules Engine.

Example prompt:

Generate a product description for a portable, rechargeable hand warmer called 'WarmPal'. 
Target audience: outdoor enthusiasts and commuters in cold climates. 
Tone: friendly and informative. 
Voice: casual, yet knowledgeable. 
Key features to highlight: compact design, 12-hour battery life, fast-charging capability, and eco-friendly materials. 
Mention that it's perfect for hiking, commuting, or any outdoor activity in the winter. 
Include technical specifications like dimensions (3x2 inches), weight (4 ounces), and material (recycled plastic). 
Avoid technical jargon and make the description accessible to non-experts.

By following these guidelines and using the example prompt as a template, you can craft effective and detailed prompts that will help generate product content tailored to your needs and the preferences of your target audience.

Frequently Asked Questions

Q: Can the AI combine text attributes and images in a single generation?

You can choose to generate content either from attributes or from assets (optionally including attribute context), but not both at once in a unified prompt.

Q: What happens if I have too many or very large assets?

Only the first 5 supported images and 1 PDF will be considered (size limits apply; larger assets are ignored).

Q: Is my data safe?

Yes. Akeneo uses OpenAI’s Enterprise API, which does not use your data for model training and does not store your data after processing.