Terms for the Automated Assistant
Draft of the next revision. The effective date will be set upon approval. This revision is unpublished and does not replace the current documents.
These terms apply to Akmelo functions that use automated intelligent tools to prepare instructions, code, transformations, and explanations.
1. Purpose and limitations
The assistant makes suggestions based on available context and tool results. Its answers may contain errors, incomplete information, incorrect interpretations of page structure, or code that does not work. An answer is not a legal opinion and does not establish that collection is lawful. A prepared scenario does not prove that the result is complete, accurate, or permitted.
Before running a task or using its result, the user checks the purpose, sources, permissions, access settings, fields, recipients, and possible consequences. Automatic execution within the authority granted by the user remains use of the Service at the user's direction. Material financial, legal, and other decisions require independent verification of the facts by a competent person.
2. Information that must not be submitted
Do not include personal data, passwords, tokens, keys, payment secrets, legally protected secrets, or another person's confidential information in messages, attachments, examples, or materials provided to the assistant. Use fictional or properly anonymized values in examples that cannot be linked to an individual. Replacing a name alone may be insufficient if identifying context remains.
Tools that remove sensitive information do not replace the duty to comply with these restrictions and do not guarantee detection of all personal data. If access credentials are needed for an ordinary task, use only the Service mechanism intended for them and stay within your lawful authority. This does not permit placing secrets in chat text or code.
3. Processing and rights
Conversations with the assistant on the public website are stored in the Service, including original visitor messages, replies, and conversation titles. A conversation can be resumed from the same browser with the functional cookie retained, without creating an account. Before sending a request, the visitor is shown a history notice and a link to the Personal Data Processing Policy. Preparing context for an external intelligent tool includes automated filtering, which does not guarantee removal of all personal data or anonymise the stored original conversation.
Removing a conversation from visible history ends the visitor's access to it but does not destroy server records. The purposes and periods of retention, and the procedure for requesting cessation of processing or destruction of personal data, are governed by the Personal Data Processing Policy. Starting a new conversation does not erase the previous one retained under those conditions. Do not use public conversations as a secure store for confidential material.
The context needed to provide the function may be processed by external providers. The Service aims not to add internal account identifiers or payment data to a request. Text, code, and attachments selected by the user for processing may nevertheless contain personal data. Complete anonymization of free-form content is not guaranteed. These terms do not give providers an independent right to disclose user data or replace legally required information about processing. The legal bases, categories of recipients, and necessary conditions are set out in the Personal Data Processing Policy and the applicable agreement.
Users must have the rights required to submit their materials. Generated results may resemble other materials, contain protected elements belonging to others, or lack independent legal protection. The Service does not promise that results are exclusive or grant rights that belong to third parties. Users check the required permissions before publishing or distributing results.
4. Charges, errors, and prohibited uses
Assistant use consumes resources under the applicable plan and settings. An unsuccessful generation does not automatically mean that no resources were used; disputed charges are reviewed under the payment terms. Do not use the assistant for activities prohibited by the Acceptable Use Policy or to circumvent Service restrictions.
Mandatory security review of cloud task code, including intelligent risk assessment, is paid for by the Provider. The user is not charged for that review, and it does not consume the user's automated assistant quotas or traffic allowances. It is a separate check under the Acceptable Use Policy rather than generation requested by the user. Work performed by the task itself in the cloud before a protective stop is accounted for under the ordinary billing terms; disputed usage and refunds are handled under the payment terms and law.
Review results may be retained for reuse with regard to the code revision and review rules. An automated signal can hold a particular run, but staff confirm violations and consider appeals. Rights to raise an enquiry and correct an error remain available.
If sensitive information is included by mistake, stop sending it and contact support@insoweb.ru. For an execution error, contact support@insoweb.ru. Provide the identifiers needed to investigate without repeating the secrets. The Provider assesses containment and deletion measures in light of the actual processing and the law. These terms do not exclude the Provider's liability for its own violations.