AI FAQ

Key points at a glance

Our AI functionalities are operated exclusively on our own servers in Germany. The following principles apply:

  • Data Usage: Only the information you actively provide is processed, and only to the extent necessary for the requested calculation. No automatic or hidden data transfer takes place.
  • Data Protection & Security: All data is transmitted and stored in encrypted form, is fully GDPR-compliant, and is not shared with third parties.
  • Training: Your data is not used for training AI models. It is used solely for the calculations you explicitly initiate.

1. Where is AI used in the tool?

2. What data is transmitted to the AI?

3. Is my data automatically sent to the AI?

4. Where is the AI hosted and how is data transmitted?

5. How secure is my data when using the AI?

6. Is my data being used to train the AI?

7. Which models are used?

8. Can I review or adjust the AI results?

9. How does the AI handle missing or incomplete data?

10. Are the results verified, or can the AI make mistakes?

11. What are the advantages of AI-based calculations compared to manual processing?


1. Where is AI used in the tool?

Artificial intelligence is currently used in two areas of the tool to help calculate carbon footprints:

Scope 3.1 & 3.2:

  • The AI supports the calculation of emissions for purchased goods and services (Scope 3.1) as well as capital goods (Scope 3.2) as part of creating your Corporate Carbon Footprint (CCF). The AI can perform calculations using either the spend-based method or the activity-based method:
    • Spend-Based: Emissions are estimated based on the amount of money spent on goods or services.
    • Activity-Based: Emissions are calculated using actual activity data, such as quantities, weights, or units of use.
  • For this purpose, the AI processes the information you provide (e.g., commodity, supplier, country) and automatically identifies suitable emission factors.

PCF (Product Carbon Footprint):

  • Our AI supports you in the automated creation of product carbon footprints (PCFs). All relevant life cycle phases up to the factory gate are taken into account (cradle-to-gate).
    • An AI-powered scraping service analyzes the product URLs you provide to extract relevant product information.
    • Based on the provided/identified product information, the AI can detect components, materials, manufacturing processes, transportation, and packaging, and match them with relevant emission factors.
    • If you additionally provide a Bill of Materials (BOM), the AI will process this data as well to determine emission factors with greater accuracy.

2. What data is transmitted to the AI?

Only the information necessary for the respective calculation and actively provided by you is transmitted to the AI:

Scope 3.1 & 3.2:

  • The AI only receives the item name, commodity, supplier, country, and the chosen calculation method.
  • It does not receive prices, weights, or quantities. In addition, all data is transmitted in an anonymized form, meaning the AI has no knowledge of our clients’ identities.

PCF (Product Carbon Footprint):

  • The AI processes the product information you provide, which may include: product name, category, description, weight, dimensions, volume, manufacturer name and country, brand, components, materials, process steps, and – if supplied – the bill of materials (BOM).

3. Is my data automatically sent to the AI?

No. Data is only transmitted to the AI when you actively execute one of these AI-supported functions:

  • Starting a Scope 3.1 / 3.2 emissions calculation
  • Generating a PCF calculation.

Without your explicit action, no data is shared with the AI and no automatic processing takes place.


4. Where is the AI hosted and how is data transmitted?

Our AI solutions for emissions calculations are hosted exclusively on our own servers located in Germany. All data transmissions are encrypted (“in transit”, via HTTPS and VPN using WireGuard at the kernel level), and multiple redundancies and backups ensure additional reliability. Your data never leaves Germany and is protected by comprehensive technical and organizational measures.

To identify relevant product information on publicly accessible websites (URLs), we use the external scraping service Firecrawl. Here as well, only the URLs you actively provide are processed.


5. How secure is my data when using the AI?

The processing of our AI services takes place exclusively on our own servers in Germany. All data used within the AI services is transmitted and stored in encrypted form (“end-to-end encryption”), is fully GDPR-compliant, and is not shared with any third parties.

For more detailed information on data security, please visit our Trust Center.


6. Is my data being used to train the AI?

No, your data is not used to train our AI models. It is only used to execute the function you request, and once the calculation is complete, it is not processed any further.


7. Which models are used?

We use only open-weight models that have been trained in the EU or the USA. Open-weight means that the trained model parameters are openly accessible, allowing us to operate these models entirely within our own secure infrastructure. As a result, your data always remains within our system, is not shared with third parties, and is not used to further train the models. We exclusively rely on models developed and trained in the EU or the USA, such as Gemini or OpenAI. Models from Asia are not used.


8. Can I review or adjust the AI results?

Yes. All results generated by the AI are presented transparently. In addition, we provide a confidence score indicating how certain the AI is in its recognition or classification as a well as a reasoning explaining the choices. You always have the option to review, validate, and manually adjust the results, ensuring that you retain full control at all times.


9. How does the AI handle missing or incomplete data?

Our AI can work with both complete and incomplete information. If certain data is missing, the AI can generate estimates based on the information available. For each estimate, the AI provides a confidence score that indicates the reliability of the assumption, as well as a short reasoning explaining the basis of the estimate or choices.

Importantly, whenever you provide primary data on materials, quantities, or emission values, these will always take priority. Any automatically generated estimates can be replaced by your own more precise inputs at any time.


10. Are the results verified, or can the AI make mistakes?

We continuously improve our AI and ensure compliance with the highest quality standards as well as established standards for emissions calculation. However, as with any automated process, errors may occur. We therefore recommend that you review the results and share feedback with us if you notice any inconsistencies or have suggestions for improvements.


11. What are the advantages of AI-based calculations compared to manual processing?

AI-powered functions offer several key benefits:

  • Time savings: Automated analysis reduces repetitive manual effort and speeds up calculations.
  • Scalability: Even large or complex datasets can be processed consistently and quickly.
  • Reduced error risk: Standardized processes help lower the risk of input or transfer errors.
  • Transparency: Results are accompanied by confidence scores, and any estimates or uncertainties are clearly highlighted.
Did this answer your question? Thanks for the feedback There was a problem submitting your feedback. Please try again later.