Product terms · 07
Terms of Use for the Artificial Intelligence and Voice Features
NexToCloud · NTC CAPITAL HOLDING S.A.S. · NIT (Colombian tax ID) 902.096.121-7 Version 2.0 · Effective as of: October 20, 2026 · Last updated: October 20, 2026
1. Definitions#
- AI Features: the Platform features that use artificial intelligence models, such as text or voice conversational agents, message interpretation, data extraction, classification, transcription, summarization or reply suggestions.
- AI Agent: an AI Feature that interacts with End Users and performs operational actions within the configured flows (for example, confirming or rescheduling an appointment).
- Inputs: the information supplied to an AI Feature (messages, audio, instructions, context data).
- Outputs: the result generated by an AI Feature (responses, transcriptions, extracted data, classifications, summaries).
- Model Provider: the third party that provides the AI model or AI infrastructure that NexToCloud integrates, as set out in the Sub-processor List.
2. Activation and Customer control#
2.1. AI Features that interact with End Users or perform actions are activated in production only with the Customer's authorization, given in writing or through express configuration on the Platform.
2.2. The Customer retains control at all times: it reviews and approves the flows, base messages and permitted actions; it may pause or deactivate the AI Agent; and it has access to the log of conversations and actions for verification.
2.3. AI Features offered as "trial" or "beta" are governed by Clause 4.7 of the Terms.
3. Nature and limitations of AI#
3.1. AI Features are support tools. They are based on probabilistic models that may produce incorrect, incomplete or unexpected Outputs, and may misinterpret a message or a call.
3.2. NexToCloud applies reasonable accuracy and security safeguards (narrowly scoped instructions, server-side validation of actions, scope limits, escalation to human staff and monitoring), but does not guarantee the accuracy of all Outputs or the correct interpretation of all messages.
3.3. The AI Agent's actions are limited to those configured. Authority over the data resides in the Platform's server: the models have no direct read or write access to the database and only perform actions that the Platform validates.
3.4. The Customer must review the Outputs before basing relevant decisions on them, and must establish the human controls it considers necessary for its operations.
4. Prohibited uses#
In addition to the provisions of the Acceptable Use Policy, the Customer will not use the AI Features to:
a) Issue diagnoses, opinions or recommendations of a medical, legal, financial or any other professional nature. b) Make decisions with legal or similarly significant effects on individuals (for example, denying a service, credit scoring or selecting personnel) based solely on automated processing, without human review. c) Build profiles of individuals based on Sensitive Data. d) Impersonate real people or clone voices without express authorization. e) Conceal from End Users that they are interacting with an automated system.
5. Transparency with End Users#
5.1. The Customer must inform its End Users, in its privacy notice and in the channel itself, that they may be served by an automated artificial intelligence agent. The Platform identifies its agents as AI assistants.
5.2. End Users must be able to request human assistance. Flows must provide for escalation to the Customer's staff during business hours or for instructions on how to contact them.
5.3. Voice calls. When the Customer uses voice agents or call recording, the call must begin with a notice that it is being handled by an automated assistant and that it may be recorded and transcribed to handle the request. The Customer must have the corresponding authorization from the Data Subjects.
6. Data, Inputs and Outputs#
6.1. Ownership. Inputs provided by the Customer and Outputs generated for the Customer form part of the Customer Data and belong to the Customer. This does not include the System Configuration (prompts, instructions, flows and parameters created by NexToCloud), which is owned by NexToCloud.
6.2. No training. NexToCloud does not use Customer Data, Inputs or Outputs to train general-purpose AI models. Access to the models is provided through Vercel AI Gateway, configured in zero data retention (ZDR) mode: requests are routed only to Model Providers that have zero-retention agreements with Vercel and that do not use the data to train their models.
6.3. Minimization and retention. Only the information necessary for the task is sent to the models. Inputs are processed to generate the Output, and neither the AI gateway nor the Model Providers retain them after the request is completed, except for the security exceptions that some providers maintain to detect abuse, in accordance with their zero-retention policies. The conversation log is retained only in the Customer's dedicated database.
6.4. Service improvement. NexToCloud may review, with restricted access and under confidentiality, samples of conversations in the Customer Workspace solely to provide support, investigate errors or improve the Customer's own configuration, in its capacity as data processor (Encargado del Tratamiento under Law 1581 of 2012).
6.5. The principles of External Circular 002 of 2024 of the Superintendence of Industry and Commerce (SIC) (suitability, necessity, proportionality, transparency, human oversight and security) guide the design and operation of the AI Features.
7. Model Providers#
7.1. NexToCloud may use, combine or replace Model Providers and model versions to maintain the quality, cost or security of the service, in accordance with the procedure set out in the Sub-processor List.
7.2. The availability of Model Providers is outside NexToCloud's control. Their failures do not count toward the SLA, without prejudice to NexToCloud's efforts to provide fallback mechanisms or escalation to human assistance.
8. Pricing and Usage Charges#
AI and voice features may be charged through a base fee and/or Usage Charges (for example, conversations, voice minutes or processing operations), as set out in the Order Form. Usage Charges are measured using the Platform's and the providers' records. Telephony costs are governed by the Third-Party Services arrangements set out in the Terms.
9. Regulatory developments#
Colombia is currently debating a legal framework for artificial intelligence. NexToCloud will update these terms and its practices to comply with any regulation that is issued, in accordance with the amendment procedure set out in the Terms.
Versioned document. Previous versions are retained at nextocloud.ai/legal.