IT-Expertiza has released version 0.16 of its Cognitum platform, designed for corporate AI integration. The update introduces tools for tracking AI-related expenses, managing usage limits, and distributing requests across various language models.
The platform allows administrators to set daily, weekly, or monthly quotas for departments and employees. If a limit is reached, the system can either reject the request or automatically reroute it to a different model based on predefined configurations.
Cognitum provides detailed reporting on costs categorized by department, user, model, and task type. Employees can monitor their remaining usage limits directly within the interface, while IT departments gain visibility into resource utilization.
The update includes advanced routing capabilities, allowing models to be grouped into pools. The system manages traffic based on task requirements and model availability, including options to restrict sensitive data processing to local, on-premise models.
New features enhance knowledge base management, including semantic document search and automated summarization of long chat sessions. The platform supports the Model Context Protocol for integrating AI agents with corporate services.
Cognitum is deployed on-premise within the client's infrastructure, enabling offline operation. These management and security rules apply to both direct user interactions and applications connected via API.
