Arize AI

29 reviews

Arize’s platform can test data distribution changes across millions of prediction facets, pinpointing specific problems so teams can triage why models are drifting from their intended purpose.

Velvet

0 reviews

Velvet was an AI gateway designed to help engineers analyze, evaluate, and monitor AI-powered features in production environments. By acting as a proxy, Velvet warehoused every request from AI models like OpenAI and Anthropic to a PostgreSQL database, enabling comprehensive analysis and optimization of AI applications. Key Features and Functionality: - AI-First SQL Editor: Allowed users to write complex SQL queries using natural language, facilitating easy data analysis. - Collaborative Data Tools: Enabled teams to turn queries into tables, graphs, and alerts, promoting shared insights and collaborative decision-making. - Real-Time Data Utilization: Provided interoperable API endpoints to build analytics and product features using live data. - Model Evaluations: Offered frameworks to run experiments on request logs, testing models, settings, and metrics to ensure AI features function as expected. - Data Retention and Archival: Implemented policies to maintain performance, reduce storage costs, and provide easy access to historical data for analysis. Primary Value and User Solutions: Velvet addressed the challenges of managing and optimizing AI features in production by providing tools for comprehensive data analysis, real-time monitoring, and collaborative development. It enabled product teams to understand usage patterns, troubleshoot issues, calculate costs, and evaluate models effectively. By warehousing AI requests and offering intuitive querying capabilities, Velvet empowered users to build more reliable and efficient AI applications. In 2025, Velvet was acquired by Arize, an enterprise platform specializing in AI evaluation and observability, to further enhance developer-first AI infrastructure.

Arize Phoenix

0 reviews

Phoenix helps you understand and improve AI applications by giving you a workflow for debugging and iteration. You can send detailed logging information, known as traces, from your app to see exactly what happened during a run, score outputs using evaluation tests to identify failures and regressions, iterate on your prompts using real production examples, and optimize your app with experiments that compare changes on the same inputs. Together, these tools help you move from inspecting individual runs to improving quality with evidence.