A: elvex is a management layer that serves as an interface between users and AI providers such as OpenAI, Anthropic, Cohere, and others. It is not a Large Language Model (LLM) in itself but rather enhances the capabilities of the AI providers.

A: elvex acts as a mediator between users and AI providers. Users interact with elvex to make requests and receive responses generated by the underlying AI provider. elvex manages the communication, authentication, and security aspects, providing a seamless experience while leveraging the power of the AI providers.

A: No, elvex does not possess independent models. It relies on the language models and algorithms provided by AI providers. These providers deliver the actual AI capabilities and are responsible for training, fine-tuning, and maintaining their models.

A: Yes, the same hallucination risk may exist when using elvex as when using the AI providers directly. It is important to understand that elvex enhances the access to AI providers but does not independently control the content generated by them.

When elvex apps are paired with Datasources, the app’s knowledge set is significantly confined to the information in the Datasource, lowering the risk of hallucinations considerably. Responses from apps with connected Datasources will provide citations so you can easily reference the source material and confirm the accuracy of the response.

A: To minimize the occurrence of hallucination, it is crucial to provide clear instructions and prompts through elvex. However, it is recommended to consult the guidance and recommendations provided by the AI providers themselves on how to handle the risk of hallucination for each platform.

Also, see above re: Datasources.

A: elvex can implement filters, constraints, or guidelines to help reduce the risk of hallucination, depending on the AI provider's capabilities. However, the effectiveness of such measures may depend on the specific AI provider and its AI models.

A: No, elvex, like any other AI-based system, cannot guarantee complete protection against risks such as hallucination. It is recommended to be cautious, follow best practices, and rely on the expertise and guidance of the AI providers to address potential risks.

A: To understand the risks of hallucination and how to handle them effectively, it is best to refer to the documentation, guidelines, or support provided by the respective AI providers. They will offer insights, recommendations, and best practices regarding these concerns.

A: elvex employs industry-standard encryption protocols, secure connections, and other security measures to protect user data during transmission. It may also anonymize or aggregate data to further enhance privacy. However, it is important to review elvex's privacy policy and terms of service for specific details on data handling and security practices.

A: Users should exercise caution when sharing sensitive or confidential information while using elvex or any AI-powered service. It is recommended to avoid entering personal, financial, or sensitive data unless necessary. Additionally, it is advisable to review the privacy policies and terms of service of both elvex and the AI providers to understand how they handle and protect user data. When in doubt, refer to your company's guidelines and policies.

A: No, elvex does not use prompt, response, or uploaded data sources to train models. As a management layer for AI models, elvex does not have its own models and, therefore, does not utilize user data for training purposes. elvex acts as a facilitator between users and AI providers, ensuring secure communication and deployment of AI models without directly accessing or using the data for training. The responsibility for training models lies with the AI providers who have their own data and training processes. If you are concerned about your data being used to train models operated by the AI providers, we recommend reviewing the terms of service for those providers.

A: As of July 17, 2024, elvex supports connections with OpenAI, Azure OpenAI,Anthropic, Cohere, and Mistral, though each customer must authorize and activate the platforms and models they wish to make available to its users. If there is a platform or model that you would like to connect to elvex which is not currently supported, please email us at [email protected]. Our goal is to support all major platforms and models across all modalities and your feedback will help us prioritize which platforms we should support next.

A: The response times for conversations with AI platforms through elvex are not expected to have any noticeable differences when compared with using the platforms directly. There can always be slight variations and it is important to note that actual response times may depend on factors such as the complexity of queries, the specific AI platform being used, and the availability of that platform's resources. However, elvex aims to offer a seamless experience with response times comparable to those experienced when directly using the AI platforms themselves.

A: Yes, elvex supports Single Sign-On (SSO) and access management through Google-based sign-in or Okta. Additionally, elvex uses Okta's Auth0 for authentication.

A: elvex serves as a unified interface that facilitates communication between users and various AI platforms. Customers can access the AI platforms through elvex by providing their credentials for each respective platform. It is not required for each user of elvex to also have a user account with the AI platforms directy.

A: No, elvex does not charge any additional fees for accessing AI platforms. elvex itself does not impose any charges beyond its own service fees. However, it is important to note that customers will still be responsible for any costs associated with their usage of the AI platforms, as determined by their agreements with those platforms. elvex simply provides a platform for users to conveniently access and interact with AI services, using their personal credentials, without any additional charges directly related to the AI platform usage.