About AI product classification
AI product classification analyzes your product category and description and suggests product tax codes, helping you map your products faster and more accurately.
Why use AI product classification
Accurate product tax codes are essential for correct tax calculation. Manually assigning product tax codes can be time-consuming and error-prone, especially when you have a large product catalog. The AI product classification tool analyzes your product information and suggests the most appropriate product tax codes based on product descriptions, categories, and industry standards.
- Faster classification
- Get product tax code suggestions in seconds instead of spending time researching codes manually.
- Improved accuracy
- AI suggestions are based on extensive tax code databases and product categorization standards.
- Confidence indicators
- Each suggestion includes a confidence level so you can make informed decisions.
- Alternative options
- Review the other tax code suggestions ranked by relevance to find the best match.
Note:
To opt your organization out of AI product classification, create a support case.
How it works
When you submit a product for classification, the AI evaluates your product description, category, and any additional information you provide. It then returns up to three tax code suggestions, each with a confidence level that indicates how closely the suggestion matches your product. You can review these suggestions, compare alternatives, and select the product tax code that best fits your product.
| Confidence level | Description |
|---|---|
| High | The AI has strong confidence that this tax code is correct for your product. |
| Medium | The AI has moderate confidence. Review the suggestion thoroughly to confirm it matches your product. |
| Low | The AI has limited confidence. Consider reviewing the alternative suggestions. |
| Very Low | The AI has minimal confidence. You should provide more product details. |
CAUTION:
Always review suggested product tax codes before you map them to your products. A high confidence score doesn't guarantee the code is correct for your product.
Tips for better classification results
- Provide detailed descriptions
- Include specific details about your product's purpose, materials, and intended use.
- Select the correct product or service type
- This helps the AI narrow down the appropriate tax codes.
- Choose an accurate category
- Selecting the right category improves suggestion relevance.
- Use additional fields
- When available, enter HS codes, CN8 codes, or UNSPSC codes to provide more context.
Data privacy and security FAQ
- How does the AI product code mapping tool use customer data?
- The AI product code mapping tool processes the product data you submit for classification, including product descriptions, categories, and any additional information you provide. This data is sent to the AI model solely to generate tax classification suggestions (such as Sovos product tax codes and HS codes). The tool also queries a Sovos-curated knowledge base of tax classification entries to ground its suggestions in authoritative data. Your product data is used only for the duration of the classification request and is not stored or retained by the AI after processing.
- Which LLMs are being used?
- The AI product code mapping tool is powered by large language models (LLMs) accessed through AWS Bedrock, Amazon's managed AI service. AWS Bedrock provides enterprise-grade security, compliance, and performance while giving Sovos access to leading foundation models.
- Do any of the LLM providers store customer data?
- No. Sovos has restrictive agreements with LLM providers that prohibit them from storing, using, or training on your product data. The LLM providers process each classification request individually and securely. They send their responses directly back to Sovos and don't retain any information.
- Can I limit or restrict the data that is shared with the AI product code mapping tool?
- The AI product code mapping tool only processes the product data you explicitly submit for classification: product descriptions, categories, product-or-service type, and any optional fields such as additional information or existing tax codes (HS, CN8, UNSPSC). It does not access any other business data, transactions, or account information beyond what you include in the classification request. You control exactly what product information is submitted.
- Does the AI product code mapping tool use my inputs and outputs to train Sovos products and services?
- No. The AI product code mapping tool doesn't use your inputs or outputs to train Sovos products and services. Sovos only uses anonymized interaction data to improve the overall experience. This data might include feature usage patterns and feedback that you choose to provide. The content of your classification requests is not used for training.
- Is any data transferred outside the current site for AI processing?
- Yes. The AI product code mapping tool sends your product data to AWS Bedrock through encrypted SSL/TLS for both knowledge base retrieval (vector search) and LLM inference, and returns the classification results. Only the product data you submit for classification is transferred. No other business data, transactions, or account information is transferred.
- How does the AI product code mapping tool ensure that my data is protected?
- In addition to the restrictive policies Sovos maintains with LLM providers, the AI product code mapping tool continues to adhere to the existing security practices in place, including JWT-based authentication (ES384) and permission-based authorization for all classification endpoints. All API requests require a valid authentication token with the appropriate permissions. For more detailed information on how LLM providers secure their platforms, visit the sub-processors list.
- Does the AI product code mapping tool impact my compliance with GDPR?
- No. Sovos is committed to helping you comply with GDPR and local data protection requirements. The AI product code mapping tool processes and transmits data in accordance with the Sovos Privacy Policy, Data Processing Addendum, and GDPR commitments.