General Questions
1. What is the AI Answers feature and how does it work?
AI Answers use Generative AI to generate answers to questions or searches in Cleary based on the information available in the Cleary platform.
When added to a Slack channel, the Cleary AI Slack bot can automatically detect when a user is asking a question and will provide relevant answers from the knowledge base. The bot can also create FAQs from questions as they get answered by other users, and keep those FAQs up to date.
For example, if a user asks about the upcoming company holiday schedule, and there is no documented knowledge about it, the bot won’t answer. When another user answers with the specific schedule, the bot will create a relevant FAQ in the closest matching Page, such as “Company Holidays” with the answer provided from the user.
2. How accurate are the AI-generated answers?
The AI aims to provide accurate answers based on the data it has access to. However, it’s important to double-check the information as AI can sometimes provide incorrect or outdated answers.
3. What happens if the AI provides an incorrect answer?
Our product messaging encourages users to double-check AI-generated answers for critical information. We also provide a mechanism for users to give feedback on answers where they can report if something was incorrect. This feedback is accessible by Super Admins and can thus be used to update your company’s content in Cleary to improve AI Answers to similar questions in the future.
Data Privacy and Security
1. Does Cleary use any third-party AI models?
Cleary utilizes multiple third-party AI models, including AWS Bedrock, Anthropic, and OpenAI. Different LLM models from these vendors are used for various use cases to provide the best possible service.
AI vendors and models may change over time as we strive to improve our services. We regularly evaluate our model providers and may use models from Cohere, Groq, or other open source models in the future as we strive to improve our services. All third-party AI model vendors are evaluated for security and privacy before being used and are SOC 2 compliant. For detailed information, please refer to our Data Sub-processors page where you can subscribe to updates.
2. Is my data stored by third-party AI models?
AI model vendors may store data for a limited period of time, generally 30 days or less. Data is used solely for generating the necessary responses and then deleted as per their data retention policies. You can read about those below.
3. Is my data used to train or fine-tune third-party AI models?
No, your data is not used to train, fine-tune, or improve third party AI models.