Why are we updating our Data Processing Agreement (DPA) and Privacy Policy?
We are continuously developing Candis to create features that make the day-to-day work of accounting and finance teams easier—through better automation, easier access to information, and reliable support.
In doing so, it is important to us to implement new features in compliance with applicable data protection requirements from the very beginning and, to the extent possible, to rely on European service providers. That is why we are updating both our Data Processing Agreement (DPA) and our Privacy Policy to transparently reflect our current features and the service providers we use.
The most significant changes relate to the future mandatory use of Intercom for our customer support and the expansion of our AI-powered features. In addition, we are adding two technical service providers that may be used for specific Candis features in the future.
We are aware that every additional subcontractor adds complexity. We therefore do not make such decisions lightly. At the same time, we do not intend for the number of our subcontractors to increase indefinitely: Over the course of 2027, we will gradually reduce our use of Gini for invoice data extraction and eventually phase it out.
Overview of Changes to the Privacy Policy and General Terms and Conditions
Privacy Policy:
Description of new AI features
Intercom is categorized as technically necessary
Changes to the General Terms and Conditions:
New subcontractors: Intercom, Bunny, Clevercloud
Clarification of clauses and removal of inapplicable clauses
1.1 Revision of data categories for better clarity
3.9 regarding the provision of data carriers, as this does not occur in practice; clauses 3.11 and 4.4 have been deleted accordingly.)
7.1 and 7.2: Clearer specification of a four-week deadline for engaging new subcontractors and clarification of the process
The changes will take effect on November 15, 2026.
Why will the use of Intercom be mandatory in the future?
Intercom is our central tool for support and communication with Candis users. Until now, the use of Intercom has been optional; in the future, Intercom will be an integral part of the regular use of Candis.
This ensures that we can continue to provide our customers with the support they need to use Candis and handle support requests reliably and consistently. The processing is carried out to provide the contractually agreed-upon support services based on Article 6(1)(b) of the GDPR.
What personal data is processed via Intercom?
In the context of support, the following data in particular may be processed:
Name and business email address,
Organization or Candis account,
Content of support requests and messages,
Technical information necessary to process a request,
Information regarding the use of Candis, to the extent it is needed to process a support request.
Which data is actually processed depends on the specific support request.
Is the purpose of Intercom changing?
No. Intercom will continue to be used for customer service, support, and related communication with our users.
The change is that, going forward, Intercom will no longer be activated only optionally but will be an integral part of the regular use of Candis.
Where is the data processed by Intercom?
For our Intercom workspace, we use the European data region. Customer data processed via Intercom is generally hosted and processed within the European Union on Amazon Web Services servers in Dublin, Ireland.
In limited cases, such as in connection with certain support or subcontracted processing services, processing outside the European Economic Area cannot be completely ruled out. Transfers to the U.S. are carried out in accordance with Art. 44 et seq. of the GDPR. Intercom, Inc. is certified under the EU-U.S. Data Privacy Framework; in addition, the European Commission’s Standard Contractual Clauses may apply.
Why was the privacy policy updated regarding AI?
We have expanded the scope of our AI-powered features and would like to provide greater transparency regarding the processing steps for which we use Google Cloud Vertex AI and AWS Bedrock (hereinafter referred to as Google Cloud and AWS).
The changes specifically concern more advanced data extraction and automation, the linking of different types of information, and the Candis AI assistant Cal.
What new or expanded AI features are available?
The expanded features include, in particular:
the extraction of additional information from specific document types, such as the names of individuals listed on entertainment expense receipts,
the analysis and extraction of relevant information from email texts in the Candis email inbox,
the automatic recognition of relationships and the linking of related content, such as invoices, contracts, credit card transactions, expense reports, purchase requisitions, or other documents;
the derivation and suggestion of account assignment and classification information, such as for cost centers, general ledger accounts, teams, or approvers;
support for automated invoice approvals based on predefined criteria and rules;
the provision of the Candis AI assistant Cal.
The relevant criteria for this can be viewed or configured by authorized users of the respective customer.
What personal data may be processed in this context?
Depending on the function, personal data that is already part of the documents and processes existing in Candis may be processed, for example:
Names of contact persons, employees, signatories, or other individuals named on documents,
names of guests,
email addresses and phone numbers,
bank and payment information, in particular IBAN,
the content of invoices, contracts, emails, and other documents,
information on credit card transactions and expenses,
user, approval, and editing information,
other personal information contained in the analyzed content.
The models used process only the data transmitted for the purpose of providing the respective Candis feature.
Why does Candis use both Google Cloud Vertex AI and AWS Bedrock?
Depending on the specific use case, the two platforms can be used alternatively or in combination to ensure high availability, reliability, and quality of the processing results.
Where is the data processed for the AI features?
Processing via Google Cloud Vertex AI takes place in the European Google Cloud region; processing via AWS Bedrock takes place in the European AWS region. Data for Cal is processed exclusively within the European Union.
Is customer data used for the general training of AI models?
No. Processing is carried out exclusively to provide the respective Candis feature.
The data is not used for general training or to improve generally available AI models. Furthermore, data transmitted to AWS Bedrock is not made available to the providers of the models hosted there for the purpose of training their models.
What is Cal?
Cal is an AI-powered chat assistant that can be used within Candis as well as via connected applications such as Microsoft Teams and Slack. For example, users can ask Cal questions in natural language about invoices, receipts, and transactions, retrieve information about pending tasks and approvals, and receive reminders and notifications regarding these processes.
Cal processes only information from those companies in Candis for which the respective user has the appropriate authorization.
What AI is used for Cal?
We use AWS Bedrock for Cal. Through AWS Bedrock, Cal utilizes a Claude model from Anthropic.
The Claude model processes the information required for each query within the AWS environment used by Candis. This data is not transmitted to Anthropic, and Anthropic does not have access to customer data processed via Cal.
The data is processed exclusively for the purpose of providing Cal and for responding to or executing the respective user request. It is not used to train Claude or to improve generally available AI models, nor is it used for any other purposes of the AI providers involved.
What personal data does Cal process?
Cal processes only the information necessary for the respective user request. This may include, in particular:
messages and requests entered by the user,
the user’s name and information,
information from invoices, receipts, and other documents,
information regarding transactions,
information regarding approvals, open tasks, and processing statuses,
other personal information necessary to respond to the respective request.
Which data is actually processed depends on the specific request and the user’s permissions.
When using Cal via Microsoft Teams or Slack, the content entered and output there is additionally processed within the scope of the customer’s configuration of the respective platform.
Does Cal make automated individual decisions within the meaning of Article 22 of the GDPR?
No. Cal is designed to provide information and support Candis’s existing processes.
Cal does not make any decisions based solely on automated processing regarding natural persons that have legal or similarly significant effects within the meaning of Article 22 of the GDPR.
What does automated invoice processing mean?
Candis can automate certain processing steps based on predefined criteria and rules. This may involve, for example, assigning an invoice, requesting or performing an approval, further processing steps, or an export.
The relevant criteria can be viewed or configured by authorized users of the respective customer.
Which additional subcontractors are included?
We are also adding Clever Cloud SAS (Nantes, France) and BunnyWay d.o.o. (bunny.net, Ljubljana, Slovenia) to our list of subcontractors. Both providers are used only for specific Candis functions and do not replace our existing core infrastructure.
Clever Cloud, for example, can be used to host smaller Candis functions. A current example is a Customer Dashboard that aggregates billing information from multiple companies belonging to a single customer.
The specific personal data processed depends on the respective application. In the case of the Customer Dashboard, for example, user and access data, as well as billing or supplier data linked to individuals, may be processed.
In certain cases, bunny.net can provide content delivery and edge infrastructure for the secure and high-performance delivery of content. In this context, primarily technical connection and request data are processed, such as the IP address or anonymized IP address, the URL accessed, and the user agent.
Are Clever Cloud and bunny.net automatically used for all customers?
No. Inclusion in the AVV initially means that Candis may use these providers for specific functions.
Whether personal data is actually processed via one of these providers depends on the specific Candis function being used and its technical implementation.
Can I object to the addition of new subcontractors?
Yes. Before engaging additional subcontractors, we will notify you by updating the list of contracted subcontractors. You may object in writing to changes within a reasonable period of time for good cause by contacting support@candis.io (Section 7.2 of the General Terms and Conditions).
Who can I contact with questions?
If you have questions about data protection, our Data Protection Department can be reached at datenschutz@candis.io.