Klark acts as a processor for data processed on behalf of its clients and as a controller for its own activities. Our approach includes security measures, data minimization, rights management, and the monitoring of corrective actions. Our privacy policy details our practices and current developments: https://www.klark.ai/privacy-policy.
Klark is firmly committed to data security and is SOC 2 Type 2 certified.
To achieve this, we use Vanta, a recognized tool that helps us to continuously monitor our systems and apply best security practices, thus guaranteeing the protection of our customers' data.
You can find a list of our security controls and processes on our Trust Center.
Klark retains the data necessary for the operation of the service in accordance with its intended purpose, the client's instructions, and applicable retention rules. The primary hosting is located in Europe. Some data processing by third-party providers may take place outside the European Union; their location and the regulatory framework governing them are described in our privacy policy: https://www.klark.ai/privacy-policy.
Retention periods depend on the data and its intended use. The rule for deleting pseudonymized correspondence, which is currently being implemented, stipulates that such data will be deleted upon closure of the ticket or after 30 days of inactivity if the ticket remains open. Using a suggestion does not trigger this deletion. Conversation content and logs are subject to their own retention periods, as detailed in our privacy policy: https://www.klark.ai/privacy-policy.
Your data is used solely for your use and is completely segregated from the data of our other customers. Your data is of course used to generate suggested answers and help your customer service agents improve productivity and quality, but it is not used to train our algorithms.
The information necessary to process requests is transmitted to the models used for the service. In particular, we use masking and replacement of identifiers to reduce the exposure of personal data. These mechanisms do not guarantee the removal of all identifying information, particularly in free-form text, attachments, or contextual elements. Data flows remain subject to the client’s instructions and the applicable protection and transfer safeguards. Our privacy policy details these terms: https://www.klark.ai/privacy-policy.
At Klark, we're "LLM agnostic" and don't want to be dependent on specific LLMs. Our promise is to bring you the best of AI at every moment and according to your needs. That's why we've developed an orchestration algorithm capable of calling on the best LLMs on the market.
Today, for example, we use the services of GPT 5 (OpenAI), Mistral Large and Claude 4.5 (Anthropic).