Artificial Intelligence Addendum

1.1 Purpose. This Artificial Intelligence Addendum (the "AI Addendum") governs the Customer's use of the AI Features made available by DinMo SAS ("DinMo") as part of the Service. It supplements the General Terms and Conditions of the DinMo Service, or any executed written agreement between the Customer and DinMo, together with its Order(s) and Data Processing Addendum (collectively, the "Agreement").

1.2 Acceptance. The AI Features are optional. This AI Addendum is accepted, and becomes effective, on the earlier of (a) its acceptance by a Customer administrator when enabling the AI Features in a Workspace, or (b) the first use of any AI Feature by the Customer or any of its Users. A Customer administrator who enables the AI Features represents that they are authorised to bind the Customer. If the Customer does not agree to this AI Addendum, it must not enable or use the AI Features.

1.3 Disabling. The Customer may disable the AI Features for a Workspace at any time through the Workspace settings. Disabling stops any further transmission of Inputs to AI Providers for that Workspace. AI Providers do not receive credentials for, or direct access to, the Customer’s Data Source or data warehouse. Data previously processed remains subject to the retention and deletion periods set out in Section 5.5 and Annex 1.

1.4 Precedence. This AI Addendum is a schedule to the Agreement. In the event of conflict regarding the AI Features, the following order applies: (i) the Order, for any AI-specific term expressly agreed in it; (ii) the Data Processing Addendum, for the processing of personal data; (iii) this AI Addendum; (iv) the General Terms. Everything not addressed in this AI Addendum remains governed by the Agreement. Capitalised terms not defined here have the meaning given in the Agreement.

1.5 Future AI Features. This AI Addendum applies to the AI Features available at its effective date and to any AI Feature DinMo adds during the Subscription term, without the need for a new acceptance, subject to Section 9.4.

1.6 Optional and severable. The AI Features are an optional and severable layer of the Service. The Service, including every module ordered by the Customer, can be used in full without the AI Features, and the Customer’s rights and DinMo’s obligations under the Agreement are not conditional on their availability. For the purposes of Section 3.4 of the General Terms, neither the AI Features nor the Models or AI Providers supporting them constitute core functionalities of the Service, and the automatic entitlement to replacement or successor functionality set out in that Section does not apply to them, unless expressly agreed otherwise in an Order. No service-level commitment, uptime commitment, or service credit under the Agreement applies to the AI Features.

"AI Features" means any feature of the Service that uses artificial intelligence, including large language models, to generate content, analyses, recommendations, or proposed Actions from an Input, including the conversational assistant marketed as "Dan" and any embedded natural-language assistance in the Service. AI Features do not include DinMo's predictive scoring models trained on the Customer's own data (“DinMo Intelligence”), which remain governed by the Agreement and the Documentation.

"Action" means an operation that an AI Feature proposes to perform on the Customer's Workspace at a User's request, such as creating or modifying a segment, a calculated field, a journey, or a synchronisation.

"AI Provider" means a third party whose model or hosted AI service DinMo uses to deliver the AI Features, as identified in Annex 1 (AI Providers and hosting) or in the sub-processor list referred to in the Agreement.

"Input" means any prompt, instruction, question, or other information that a User submits to an AI Feature, together with the limited context that DinMo provides to the Model to respond. That context may include relevant conversation history, Workspace Memory, Workspace metadata, and results produced by DinMo-controlled tools in accordance with Section 3.

"Model" means any artificial intelligence model that powers an AI Feature, whether developed by DinMo or provided by an AI Provider.

"Output" means any content, analysis, recommendation, chart, draft object, or proposed Action generated by an AI Feature in response to an Input.

"User" means an individual authorised by the Customer to access the Service under the Customer's account.

"Workspace" means a logically separated environment of the Customer within the Service.

"Workspace Memory" means business definitions, preferences, and other context that an AI Feature stores for a given Workspace to improve its responses in that Workspace.

"EU AI Act" means Regulation (EU) 2024/1689 of the European Parliament and of the Council laying down harmonised rules on artificial intelligence, as amended.

3.1 Within the User's permissions. An AI Feature acts on behalf of the User who invokes it and only within that User's roles and permissions in the Workspace. An AI Feature cannot access data, objects, or destinations that the invoking User cannot access.

3.2 Confirmation of Actions. An AI Feature does not create, modify, activate, or delete any object in the Workspace on its own. It proposes an Action, together with a preview, and performs it only after an authorised User has explicitly confirmed it in the Service. Where DinMo later offers the possibility to pre-authorise categories of Actions (for example, for scheduled agents), that possibility will be disabled by default and enabled only by a Customer administrator.

3.3 Traceability. The Service keeps the conversation history of each User with the AI Features, and a record of each confirmed Action with the identity of the confirming User and the date and time. These records are available to the Customer in the Service and are retained as stated in Section 5.5.

3.4 Transmission on request only. An AI Feature transmits Input to an AI Provider only when a User invokes it, or when a Customer administrator has configured a scheduled task. DinMo does not transmit Customer Data to an AI Provider in the background for any other purpose.

3.4 Controlled data access. Models and AI Providers do not connect directly to, receive credentials for, or execute queries against the Customer’s Data Source or data warehouse. To respond to a User’s request, an AI Feature may use DinMo-controlled tools. DinMo executes any required query, enforces the invoking User’s permissions, and provides the Model only with the resulting information needed to respond to the request.

3.6 Data controls, minimisation and redaction. Data-blocking and hashing rules configured by the Customer in the Service apply to the AI Features. Columns blocked by the Customer are not transmitted to an AI Provider. Results produced by DinMo-controlled tools ordinarily consist of aggregated or statistical data. Where necessary to respond to a User’s request, they may also include lists of permitted attribute values or limited row-level or user-level attribute values. Before any such result is transmitted to an AI Provider, DinMo automatically redacts direct identifiers. Remaining information may nevertheless constitute Personal Data, in which case the Data Processing Addendum applies.

3.7 Transparency. The Service identifies the AI Features as artificial intelligence, and identifies AI-generated Outputs as such, in a clear and distinguishable manner at the latest at the first interaction.

4.1 Review of Outputs. Outputs are generated by probabilistic models and may be inaccurate, incomplete, inappropriate, or biased. The Customer is responsible for reviewing each Output with appropriate human judgement before relying on it, and for every Action it confirms. DinMo does not verify Outputs.

4.2 Lawful use. The Customer is responsible for the lawfulness of its Inputs, for its use of Outputs, and for the decisions it makes on the basis of Outputs, including compliance with data protection, consumer protection, advertising, and artificial intelligence laws applicable to its business and industry.

4.3 Users. The Customer ensures that its Users are informed of this AI Addendum and have a sufficient level of AI literacy to use the AI Features appropriately, taking into account their role and the context of use.

4.4 Prohibited uses. The Customer will not, and will not permit any User or third party to, use the AI Features:

4.5 Sensitive data. The Customer will not submit to an AI Feature data that the Service is not designed to process, including payment card numbers, government identifiers, and authentication credentials. Where the Customer's Workspace contains special categories of personal data, the Customer is responsible for blocking the corresponding columns under Section 3.6 or for ensuring that it has a lawful basis for their processing by the AI Features.

4.6 Breach. A breach of this Section 4 is a material breach of the Agreement.

5.1 Customer Data. Inputs, Outputs, and Workspace Memory are Customer Data under the Agreement and are subject to the protections applicable to Customer Data. For clarity, this does not mean that all Customer Data available through a Workspace is transmitted to an AI Provider. AI Providers receive only the Inputs described in Section 2, subject to the controls and limitations set out in Sections 3.4 through 3.6. To the extent that any Input, Output, or Workspace Memory contains Personal Data, DinMo processes that Personal Data as a processor on the Customer’s documented instructions, and the Data Processing Addendum applies. The Customer’s enabling and use of an AI Feature constitutes such a documented instruction.

5.2 No training on Customer Data. DinMo does not use Inputs, Outputs, conversation history, or Workspace Memory to train, retrain, or fine-tune any Model, and does not permit any AI Provider to do so. No Customer Data is made available to another customer. Workspace Memory is scoped to the Workspace in which it was created and is not shared across Workspaces or customers.

5.3 AI Providers. DinMo uses the AI Providers identified in Annex 1 (AI Providers and hosting) to deliver the AI Features, under written terms that (a) prohibit the AI Provider from using Inputs or Outputs to train or improve its models, (b) limit the AI Provider's processing to the delivery of the service to DinMo, and (c) impose data protection obligations no less protective than those of the Data Processing Addendum. For the purposes of the Data Processing Addendum, Annex 1 (AI Providers and hosting) forms part of DinMo’s Sub-Processor List, and each AI Provider identified in Annex 1 shall be deemed to be a Sub-Processor authorised under Section 8.1 of the Data Processing Addendum, whether or not it is separately reproduced in the Sub-Processor List. The addition or replacement of an AI Provider follows the sub-processor notification and objection process of the Data Processing Addendum.

5.4 Hosting and transfers. The AI Providers process Inputs and Outputs in data centres located in the region selected by the Customer for the Workspace, among the regions DinMo makes available, as stated in Annex Annex 1 (AI Providers and hosting). The European Union is the default region. A United States region applies only where the Customer has selected it. DinMo will not route Inputs or Outputs outside the selected region without notifying the Customer under Section 5.3 and implementing the safeguards required by the Data Processing Addendum. The Customer's selection of a region outside the European Union is a documented instruction under Section 5.1, and the transfer mechanisms of the Data Processing Addendum apply to it.

5.5 Retention. The AI Provider retains Inputs and Outputs only for the period stated in Annex 1 (AI Providers and hosting). DinMo retains conversation history for a limited period of twelve (12) months from the date of the conversation, for two purposes: to display it to the Customer's Users in the Service, and to operate, support, secure, and improve the AI Features, including to review the quality of Outputs, correct errors, and adjust the instructions given to the Models. For these purposes, conversation history may be accessed by authorised DinMo personnel bound by confidentiality obligations, in accordance with the Data Processing Addendum. It is never used for the purposes excluded in Section 5.2. Conversation history is deleted at the end of that period, and Workspace Memory is deleted when the Customer disables the AI Features for that Workspace or upon termination of the Agreement, in accordance with the deletion terms of the Agreement.

5.6 Usage data. DinMo may generate aggregated and anonymised statistics on the use of the AI Features, such as volumes, latency, and error rates, which do not identify the Customer, its Users, or any data subject, as permitted by the Agreement.

6.1 Inputs and Outputs. As between the parties, the Customer retains all rights in its Inputs and owns the Outputs generated for its Workspace. To the extent DinMo holds any right in an Output, DinMo assigns it to the Customer upon generation. The Customer may use Outputs for any lawful purpose, subject to this AI Addendum.

6.2 Non-unique Outputs. Because of the nature of the Models, an AI Feature may generate identical or similar Outputs for other customers. Section 6.1 does not confer any exclusivity on such Outputs, and DinMo may independently develop products, features, or content similar to any Output.

6.3 AI Features. DinMo and its licensors retain all rights in the AI Features, the Models, the prompts, the underlying technology, and any improvements to them. Nothing in this AI Addendum transfers any such right to the Customer.

6.4 Feedback. Any suggestion or feedback the Customer provides on the AI Features may be used by DinMo without restriction, provided it contains no Customer Data.

7.1 Roles. For the purposes of the EU AI Act, DinMo is the provider of the AI Features and the Customer is the deployer. The AI Providers referred to in Annex 1 (AI Providers and hosting) are providers of the general-purpose AI models integrated in the AI Features.

7.2 Intended purpose. The AI Features are intended solely to assist the Customer's Users in understanding, segmenting, analysing, and activating the Customer's own marketing and customer data within the Service, under human supervision, as described in the Documentation. The AI Features are not intended for, and must not be used for, any purpose listed in Annex III of the EU AI Act, including the evaluation of natural persons for access to or pricing of credit, insurance, employment, education, or essential services.

7.3 Risk classification. On the basis of their intended purpose, the AI Features are not a high-risk AI system within the meaning of Article 6 of the EU AI Act. DinMo designs the AI Features to comply with the transparency obligations of Article 50 of the EU AI Act applicable to providers.

7.4 Change of intended purpose. If the Customer uses the AI Features for a purpose other than their intended purpose, or modifies them or their Outputs such that they become a high-risk AI system, the Customer is considered the provider of that system under Article 25 of the EU AI Act and assumes the corresponding obligations. DinMo has no obligation under the EU AI Act in respect of such use.

7.5 Deployer obligations. The Customer is responsible for the obligations that the EU AI Act places on deployers, including using the AI Features in accordance with the Documentation, assigning human oversight to Users with the necessary competence and authority, ensuring that the Customer Data it makes available to the AI Features is relevant and sufficiently representative for its intended use, and informing natural persons where required by law.

7.6 Cooperation. DinMo will provide the Customer, on reasonable request, with the information about the AI Features that the Customer reasonably needs to comply with its own obligations under the EU AI Act and data protection law, including a description of the AI Features' capabilities and limitations, the human oversight measures built into the Service, and the identity of the AI Providers. Each party will inform the other without undue delay of any serious incident within the meaning of the EU AI Act that it becomes aware of in connection with the AI Features.

7.7 AI literacy. Each party takes measures to ensure a sufficient level of AI literacy of its staff and other persons dealing with the AI Features on its behalf, in accordance with Article 4 of the EU AI Act. DinMo makes Documentation available for that purpose.

7.8 Data protection. Nothing in this AI Addendum authorises the Customer to make decisions based solely on automated processing within the meaning of Article 22 of Regulation (EU) 2016/679. The Customer remains responsible for the lawfulness of any profiling it performs with the AI Features and for any data protection impact assessment it is required to carry out. DinMo will assist the Customer in accordance with the Data Processing Addendum.

8.1 Evolution of the AI Features. DinMo may add, modify, replace, or withdraw AI Features, Models, or AI Providers during the Subscription term. Model changes may alter the format, style, or content of Outputs. DinMo will give the Customer reasonable advance notice of a material change through the Service or by email, and a change of AI Provider follows Section 5.3. A change that does not materially reduce the core functionalities of the Service is not a breach of the Agreement. As the AI Features are not core functionalities of the Service under Section 1.6, their addition, modification, replacement, suspension, or withdrawal shall not constitute a material diminution of the Service, a breach of Section 3.4 of the General Terms, or a right to terminate, obtain a refund, or receive any other remedy, unless an Order expressly identifies a specific AI Feature as a committed functionality and expressly states that this Section does not apply to that AI Feature.

8.2 Beta features. DinMo may make AI Features available on a beta, preview, or early-access basis and will identify them as such in the Service. Beta features are provided "as is", may be modified or withdrawn at any time, are excluded from any service-level commitment, and should not be used for production processing of sensitive data. DinMo's liability for beta features is excluded to the maximum extent permitted by law.

8.3 Suspension. DinMo may suspend or limit the Customer's access to the AI Features, or block specific Inputs or Outputs, if DinMo reasonably believes that the Customer's use breaches Section 4, an AI Provider's usage policy, or applicable law, or creates a security risk for the Service. DinMo will notify the Customer and, where practicable, give it the opportunity to cure before suspending.

8.4 Usage limits. DinMo may apply fair-use limits to the AI Features, such as a number of requests per period, as described in the Documentation or the Order.

8.5 No service levels. The availability, latency, and throughput of the AI Features depend in part on the AI Providers and may vary. The AI Features are excluded from any service-level agreement, uptime commitment, and service credit under the Agreement. Unavailability or degraded performance of the AI Features is not a breach of the Agreement and does not affect the availability measurement of the remainder of the Service.

9.1 No warranty on Outputs. Except as expressly stated in this AI Addendum, the AI Features and the Outputs are provided "as is". DinMo does not warrant that Outputs are accurate, complete, reliable, non-infringing, or fit for a particular purpose, nor that the AI Features will be uninterrupted or error-free. The warranties and disclaimers of the Agreement otherwise apply.

9.2 Liability. The limitations and exclusions of liability of the Agreement apply to the AI Features and this AI Addendum. Without limiting them, DinMo is not liable for any loss arising from the Customer's reliance on an Output, from an Action confirmed by a User, or from the Customer's use of the AI Features in breach of Section 4 or Section 7.

9.3 Indemnity. The Customer's obligation under the Agreement to defend and indemnify DinMo against third-party claims relating to Customer Data extends to claims arising from the Customer's Inputs and from the Customer's use of Outputs in breach of this AI Addendum or applicable law.

9.4 Updates to this AI Addendum. DinMo may update this AI Addendum from time to time by publishing a new version on its website and notifying the Customer through the Service or by email. Updates take effect as provided in the amendment clause of the Agreement or, where the Agreement has no such clause, thirty (30) days after notice; updates required to comply with mandatory law take effect upon publication. No update will materially diminish DinMo's obligations under this AI Addendum for the AI Features in use during the Customer's current Subscription period. The version in force is available on DinMo's website at https://www.dinmo.com/ai-addendum/

9.5 Survival. Sections 5, 6, 7.4, 9.2, and 9.3 survive the termination of the Agreement or the disabling of the AI Features.

9.6 Governing law. This AI Addendum is governed by the law and jurisdiction clauses of the Agreement.

The AI Features rely on the following AI Providers at the date of this version. DinMo may use several Models made available through those AI Providers' service. Individual Models are not listed, because a change of Model does not change the AI Provider, the processing region, or the commitments in this Annex. Any addition or replacement of an AI Provider shall be made in accordance with Section 5.3 and the notification and objection procedure set out in Section 8.2 of the Data Processing Addendum. Once effective under that procedure, the relevant AI Provider shall automatically be deemed incorporated into this Annex 1 and DinMo’s Sub-Processor List, whether or not the published versions have yet been updated. DinMo shall update those published versions accordingly.

Service: Microsoft Foundry, including Azure OpenAI. Models from several developers, all hosted and operated by Microsoft within Azure.

Processing location: Region selected by the Customer under Section 5.4: European Union by default or United States.

Training on Inputs or Outputs: None. Inputs and Outputs are not made available to the developers of the Models, including OpenAI, and are not used to train or improve any model.

Retention by the AI Provider: Up to 30 days for abuse monitoring.

Service: Google Cloud Vertex AI (Generative AI on Gemini Enterprise Agent Platform), including Gemini models developed, hosted, and operated by Google.

Processing location: Region selected by the Customer under Section 5.4: European Union by default or United States.

Training on Inputs or Outputs: None. Google does not use Inputs or Outputs to train or fine-tune any AI/ML model without DinMo’s prior permission or instruction.

Retention by the AI Provider: Google does not retain Inputs or Outputs outside DinMo’s Google Cloud account for longer than reasonably necessary to generate the Output. If Google’s automated safety tools detect suspected abuse, Google may retain the relevant flagged Inputs for up to ninety (90) days.

The usage policies of the AI Providers referred to in Section 4.4(7) are: (a) for Microsoft, the Microsoft Azure OpenAI Service Code of Conduct and the Microsoft Product Terms for Generative AI Services; and (b) for Google Cloud, the Google Cloud Acceptable Use Policy and the Google Generative AI Prohibited Use Policy, in each case as published and updated by the relevant AI Provider.