Protecting the privacy and security of the data that flows through Koquelani systems is one of our topmost concerns. We manage hundreds of kiosks, servers, and routers today using these very tools. Without trustworthy, secure systems, we believe we have no marketable product. Thus, great care and attention to how data is handled, secured, and delivered is a core focus of this product’s development.
Understanding our security / business model.
Simply put, generating revenue from your data is not what we do. We generate revenue from your usage of our service, which facilitates connectivity from your favorite AI tools to the stations you need to support.
Understanding where Koquelani sits in the privacy picture.
Koquelani’s software and services exist between your AI harness—such as Claude, ChatGPT, Grok, etc.—and your endpoints, including workstations, servers, IoT devices, kiosks, etc. While we pass your data and commands, they are fully encrypted and even we cannot see the content being transferred. Taking things a step further, in Medium and High Security modes, we also don't hold the encryption keys either, you do, further ensuring your privacy and control of the data and actions being taken.
From our perspective, this is good for you from a privacy and control standpoint, and good for us because there is no incentive for bad actors to compromise our services (we don't hold any capabilities they can exploit even if they did get in).
This makes the answer to the question, “Do we train on or use your data?” easy to answer: No. We cannot see it, so we cannot process or collect it even if we wanted to.
Understanding where your AI harness sits in the privacy picture.
This is where privacy protections become much more complicated. The landscape is evolving rapidly and is subject to change. Personal ChatGPT accounts offer a setting to opt out of model training, while OpenAI states that ChatGPT Business, Enterprise, Edu, and API data is not used for training by default. Anthropic allows Claude Free, Pro, and Max users to choose whether their data may be used for model improvement, while its commercial and API offerings are handled separately. You should review and configure the current privacy and data controls for your specific AI harness and account.
This means you must exercise absolute caution when passing sensitive data, such as passwords, private keys, security data, trade-secret data, health information, bank account information, or any other data you do not want retained or used by an AI provider.
If you require higher levels of privacy or work in an industry where sensitive information may pass through the prompt, we recommend moving to an appropriate business or enterprise account and specifically confirming its data-training, retention, and privacy settings.
If you require the highest levels of privacy, we recommend running your own AI harness on your own hardware, using tools such as AnythingLLM, LM Studio, and others. These harnesses generally operate like Claude or ChatGPT but do so on your local hardware. These setups require significantly more technical expertise, but they are available for those who require them and are fully compatible with Koquelani’s SupportFlow MCP.