Microsoft ai foundry.

Hello @Tomasz Świercz,

Yes, using Microsoft AI Foundry to automate product parameter extraction into Excel using an API key is allowed, provided it complies with the terms and technical permissions of the data source in this case, the manufacturer’s website.

This use case is considered legitimate when the AI accesses data through authorized means such as publicly available APIs or licensed data feeds. It's essential to ensure that the AI does not engage in unauthorized scraping, bypass protections, or exceed rate limits defined by the website’s robots.txt file or terms of service. Even though the task is automated through AI, the responsibility for how the data is sourced still rests with the user.

If the manufacturer provides an official API, it is strongly recommended to use that. This approach ensures legal compliance, data reliability, and avoids the risks associated with scraping. Alternatively, if an API is not available, and if permitted by the manufacturer, data can be downloaded manually and then enriched using Microsoft AI Foundry tools like Azure GPT or Assistants.

For scenarios where scraping is the only option, it must be implemented ethically and responsibly respecting usage limits, handling errors gracefully, identifying requests transparently, and ensuring that activity does not exceed what a typical user would perform manually.

Please refer this Microsoft Enterprise AI Services Code of Conduct, and Data, privacy, and security for use of models through the model catalog in Azure AI Foundry portal.

I Hope this helps. Do let me know if you have any further queries.

Thank you!

@Tomasz Świercz

You're absolutely right in noting that many manufacturers do not offer public APIs, and in such cases, using AI to process manually downloaded catalogs is generally allowed—provided those catalogs are made publicly available and downloadable by the manufacturer. Extracting product parameters from such files using AI does not fall under prohibited activity and is considered a compliant use case.

Regarding web scraping, your approach shows a good understanding of responsible data access. However, it’s important to emphasize that legality and compliance depend on the terms of use of the manufacturer’s website. Even if the scraping is performed slowly, one request at a time, and mimics human behavior, it could still be considered unauthorized if:

The site’s Terms of Service prohibit automated data collection.

The robots.txt file disallows scraping of product pages.

There are copyright or proprietary data restrictions.

To ensure full compliance, we recommend the following:

Always check the manufacturer’s website policies, including robots.txt and terms of use.

Contact the manufacturer directly to request permission or access to bulk data (via XML/CSV/Excel feeds).

If scraping is the only option, ensure your AI tool respects all ethical and technical boundaries (rate limiting, proper identification, error handling, etc.).

Thank you!

@Tomasz Świercz

Following up to see if the above suggestion was helpful. And, if you have any further query do let us know.

Thank you!