Currai

Currai enhances application development by providing comprehensive visibility and testing capabilities for conversational interfaces. Key features include: * Full tracing of prompts, tokens, and tool calls * Cost tracking per token, model, and daily usage * A/B testing of prompt versions on live traffic * Compatibility with Langfuse and OpenTelemetry SDKs * Session and user grouping for multi-turn conversations Currai eliminates the guesswork in developing applications by offering a detailed view into every interaction. Developers can easily instrument their applications with a few lines of Python or TypeScript code, gaining immediate insights into how their models receive and process information. This includes capturing inputs, outputs, model parameters for each operation, and visualizing complex request trees involving retrievers and other tools. The platform enables robust debugging and performance optimization. By centralizing all operational data, Currai allows teams to understand the exact cost implications of their conversational components and identify areas for improvement. Furthermore, its A/B testing functionality empowers developers to experiment with different prompt versions in a production environment, directly measuring their impact on quality and cost. This ensures that only the most effective and efficient components are deployed. Designed for development teams, product managers, and data scientists, Currai is ideal for anyone building and refining applications who needs to ensure reliability, manage spend, and optimize user experience. It supports confident deployment and continuous iteration, moving beyond basic debugging to proactive performance management.