The CalypsoAI Platform enables organizations to leverage the power of generative AI (GenAI) and large language models (LLMs) while moderating user interactions with the models and ensuring real-time protection against common vulnerabilities, including internal risks and external attacks.
In addition to securing organizational data, the CalypsoAI platform provides a high degree of transparency and visibility into how LLMs are being used within the organization, allowing administrators to gain valuable insights regarding user behavior and model performance.
Key Features
Scanners
The CalypsoAI Out-of-the-Box Scanners offer unique features that provide a full spectrum of secure LLM usage. Most of these scanners and their settings can be configured at both the global and group levels.
The CalypsoAI GenAI Scanner Packages provide admins the ability to block and audit arrays of prompt content that poses security or reputational risks to the organization. These scanner packages provide a range of extensive, nuanced scanners that focus on targeted functions, such as preventing personally identifiable information (PII) and prompt injection language from being included in prompts sent to external models.
The Individual Policy Scanners provide admins the ability to block, audit, or allow prompt content that could pose risks to the organization according to the context and situation. These scanner capabilities range from preventing PII, offensive or toxic language, proprietary or sensitive legal content, source code, and other content from being included in prompts submitted to external models.
The Individual Audit Scanners do not block prompts based on content, but monitor and analyze prompt content to gain insights regarding user sentiment, requests for non-business-related content, demographic biases, and named entities, such as people, competitors, or locations.
The CalypsoAI platform also enables organizations to create and deploy custom scanners tailored for their specific use cases and business needs.
Model and Provider Diversity
The CalypsoAI platform is model-, provider-, platform-, and use case-agnostic, meaning that it enables functionality across many types of models, including custom models, from multiple providers on multiple platforms for many uses. Once the models are configured (enabled) on the CalypsoAI platform, users can transition seamlessly between models, enabling them to leverage the unique capabilities of different models.
Dashboard, Prompt History, and Audit Logs
Together, the CalypsoAI platform Dashboard, Prompt History, and Audit Logs provide the Super Admin with detailed usage statistics and full traceability for every user interaction, ensuring complete transparency into user behavior and model performance over time.
Intuitive Chat Interface
The CalypsoAI chat interface provides a simple, straightforward way for users to interact with approved models. From the CalypsoAI chat interface, users can easily submit prompts, create templates to revisit valuable information, and move between models, while all activated scanners review and moderate the conversations and safeguard organizational data.
Roles and Permissions
Role- and policy-based access controls allow Super Admins to set model access and use thresholds according to the needs of individual groups.