Appendix 4 – AI Features

Appendix to the Service Agreement for Callmaker, CM Software Callmaker AB, Reg. No. 556999-9971

Applies from 2026-10-09.

1. Introduction

1.1 This appendix governs the AI Features in the Service, the parties’ roles and obligations under the AI Act, and responsibility for Output.

1.2 The appendix forms part of the Agreement and supplements the Service Agreement. In matters governed by this appendix, the appendix takes precedence over the Service Agreement. The appendix applies without a separate signature when the Order is entered into.

1.3 The processing of personal data when using the AI Features is governed by Appendix 1 – Data Processing Agreement, except for processing under section 5.5.

2. Definitions and AI Features

2.1 In addition to the definitions in the Service Agreement and Appendix 1 – Data Processing Agreement, the following apply in this appendix:

• the AI Act – Regulation (EU) 2024/1689 of the European Parliament and of the Council laying down harmonised rules on artificial intelligence, as amended from time to time.
• AI Feature – a feature in the Service that uses AI models to generate Output, such as transcription, summarisation, analysis or assessment of calls.
• Output – transcripts, summaries, analyses, scores, rankings, answers and other results generated by an AI Feature.
• Agent Assessment – Output relating to an individual User’s work performance, such as scores, ratings or rankings relative to other Users.
• Assistant Feature – an AI Feature that answers Users’ questions in the Service, such as a support assistant that answers questions about how to use the Service and an assistant that helps Users design automations (Autopilot). An Assistant Feature does not perform any actions in the Service and has no access to Customer Data beyond what the User writes in the conversation.
• the Documentation – the description of the AI Features’ intended purpose, capabilities, limitations and correct use that we provide in the Service or on callmaker.se.

2.2 The AI Features consist of Call coaching, including transcription, summarisation, content and compliance analysis of calls and Agent Assessment, and of the Assistant Features. New AI Features are covered by this appendix from the time they are made available, and we indicate in the Service or in the Documentation that they are AI Features.

2.3 An AI Feature is put into use only when you enable it in the Service. Enabling it constitutes an instruction under Appendix 1 – Data Processing Agreement. However, Assistant Features are available to your Users from the time they are made available in the Service, without separate enabling. Your Administrator can switch off an Assistant Feature for your Users at any time. Your keeping an Assistant Feature available constitutes an instruction under Appendix 1 – Data Processing Agreement.

3. Roles under the AI Act

3.1 In relation to the AI Features, we are the provider and you are the deployer within the meaning of the AI Act. Each party is responsible for fulfilling the obligations that the AI Act imposes on it.

3.2 Agent Assessment involves evaluating the work performance of persons in work-related relationships and may therefore constitute a high-risk AI system under Annex III to the AI Act. The parties shall fulfil the obligations that apply to such systems from the date on which they become applicable.

4. Your commitments

4.1 You shall:

• a) use the AI Features in accordance with the Agreement and the Documentation;
• b) ensure that Output is reviewed by a person with sufficient competence and authority before it is used as a basis for decisions affecting a natural person;
• c) not allow Agent Assessment alone to form the basis for decisions on employment, pay, promotion, work allocation, disciplinary measures or the termination of employment or an assignment;
• d) before Agent Assessment is put into use, inform the Users concerned and, where applicable, their employee representatives, and fulfil the negotiation obligations that follow from applicable employment legislation;
• e) before an AI Feature is put into use – for Assistant Features, which are made available without enabling under section 2.3, before your Users start using them – ensure that there is a legal basis for the processing and, where required, carry out a data protection impact assessment, and switch off the Assistant Features if this has not been done;
• f) not use the Service or Output to infer the emotions of Users or other natural persons, or in any other way prohibited under Article 5 of the AI Act; and
• g) notify us without undue delay if you have reason to believe that an AI Feature poses a risk to health, safety or fundamental rights, or if a serious incident has occurred.

5. Our commitments

5.1 We shall provide the Documentation and keep it up to date so that you can use the AI Features correctly and fulfil your obligations under the AI Act.

5.2 The AI Features are not designed to infer emotions, and we shall not introduce such functionality in relation to Users. Functionality that infers the emotions of other natural persons will only be introduced by amending this appendix under section 8.1, and section 8.2 does not apply to such functionality.

5.3 Call coaching runs on AI models that we operate ourselves at the Sub-processors listed in Appendix 1.2. The Assistant Features run on language models provided by the Sub-processor listed for this purpose in Appendix 1.2, with processing within the EU/EEA or Switzerland and without the data being stored by, or used to train the models of, that Sub-processor. Customer Data is not provided to third-party AI services that have not been engaged as Sub-processors under Appendix 1 – Data Processing Agreement.

5.4 We may use transcripts, Output and associated metadata to train, fine-tune and improve AI models for the Service’s AI Features, provided that the data has first been anonymised so that it cannot be attributed either to a natural person or to you. Audio files are not used for such training. Data retrieved from calendar integrations, such as Google or Microsoft accounts, is not used for such training either, whether in original, anonymised or derived form. Anonymisation takes place in accordance with Appendix 1 – Data Processing Agreement, and we shall not attempt to re-identify natural persons or you in anonymised data. Otherwise, we do not use Customer Data to train or fine-tune AI models. Anonymised data and the results of such training belong to us and may be used even after the termination of the Agreement. Such use does not constitute a breach of the section “Confidentiality” in the Service Agreement.

5.5 Calls to our support may be recorded and transcribed. We may use transcripts of such calls and the content of support cases and emails to our support to build knowledge bases for, train and improve AI Features in the Service, such as the Assistant Features, provided that the material has first been anonymised so that it cannot be attributed either to a natural person or to you. The anonymisation is reviewed manually before the material is used. Anonymised material may, as part of the knowledge base, be provided to the Sub-processor that provides the language model when an Assistant Feature answers a question. Audio files are not used for such purposes, and we shall not attempt to re-identify natural persons or you in anonymised data. We are the controller for this processing, which is described in our privacy policy at www.callmaker.se/en/privacy-policy/ and is not covered by Appendix 1 – Data Processing Agreement. Customer Data appearing in such material is anonymised in accordance with your instruction in Appendix 1.1. Anonymised data and the results of such use belong to us and may be used even after the termination of the Agreement. Such use does not constitute a breach of the section “Confidentiality” in the Service Agreement.

5.6 We retain the logs automatically generated by the AI Features to the extent and for the period required by the AI Act. We make the logs available to you on request to the extent you need them to fulfil your obligations.

5.7 If we make available an AI Feature that interacts directly with Users or with the persons you contact through the Service, or that generates audio or text sent to them, we shall design the feature so that these persons are informed that they are interacting with an AI system or that the content is AI-generated. You may not remove or conceal such information.

5.8 We may replace or update the AI models on which the AI Features are based. We will give reasonable advance notice of material changes to an AI Feature’s intended purpose or mode of operation.

5.9 Users’ questions to, and answers from, the Assistant Features are stored without any link to the User, account or customer. The content is anonymised in accordance with your instruction in Appendix 1.1 before it is used, and we may use the anonymised material to build knowledge bases for and improve the Assistant Features. Section 5.4 applies to such use, where applicable.

6. Output and liability

6.1 Output is generated automatically and is probability-based. Output may be incomplete or incorrect and does not constitute legal, employment-law or other advice. We give no warranty that Output is accurate, complete or fit for any particular purpose.

6.2 We are not liable for damage arising from Output having been used as a basis for decisions in breach of section 4.1 b) or c).

6.3 Our liability otherwise in connection with the AI Features is governed by the section “Liability and limitation of liability” in the Service Agreement and is subject to the shared liability cap there.

6.4 You shall indemnify us against claims by third parties, including authorities and your Users, arising from your use of the AI Features in breach of this appendix or the AI Act, in accordance with the section “The customer’s indemnity” in the Service Agreement.

6.5 Answers from the Assistant Features constitute Output. They do not constitute the Documentation, do not amend the Agreement and do not constitute representations about the Service’s characteristics. Automations and other configurations proposed by an Assistant Feature are Output that you shall review before they are enabled in the Service, and you are responsible for the actions that an enabled automation performs.

7. Changes to and discontinuation of AI Features

7.1 In addition to what follows from the section “Changes to and development of the service” in the Service Agreement, we may change, restrict or discontinue an AI Feature, including Agent Assessment, if in our assessment this is required to comply with the AI Act or other law, regulatory decisions or guidance from a competent authority.

7.2 We will give reasonable advance notice of such a measure, unless law or a regulatory decision requires faster action. A measure under this section does not constitute a breach of contract and does not entitle you to a reduction of Fixed Fees.

8. Amendments and survival

8.1 Amendments to this appendix are made in accordance with the section “Amendments to the Agreement” in the Service Agreement.

8.2 The addition of new AI Features under section 2.2 and updates to the Documentation do not constitute amendments to the Agreement.

8.3 Section 6 and section 4.1 b), c) and f) continue to apply after the termination of the Agreement for Output generated during the contract term.