Skip to main content
V

Velvet

The startup offers a software development platform that enables startups to authenticate users through email or SMS, onboard customers, and manage subscription payments. This infrastructure allows clients to efficiently convert high-intent users and streamline their application development process, ultimately enhancing customer accessibility and growth.

City of New York, United StatesFounded 20236300+ followers
Updated 3 months ago

Funding

$1M raised to dateRaised to date based on public sources. This may differ from the amount the company actually raised and is based only on what is publicly available on the internet.

Funding rounds are not available yet.

Founders

Product

Problem

AI-powered applications often lack robust monitoring and evaluation tools, making it difficult for engineering teams to analyze model performance, troubleshoot issues, and optimize costs in production. Existing solutions may not offer the flexibility to query logs, run experiments, and generate datasets for fine-tuning.

Solution

Velvet provides an AI gateway that enables engineering teams to analyze, evaluate, and monitor AI features in production. By warehousing LLM requests to a database controlled by the user, Velvet allows for in-depth analysis of model performance and usage. The platform facilitates the evaluation of models against specific metrics, enabling data-driven decisions on which models to deploy for each feature. Furthermore, Velvet supports continuous testing and experimentation, allowing teams to run experiments and monitor ongoing usage against predefined metrics, ensuring the reliability and efficiency of AI applications.

Target Audience

Velvet is designed for AI engineers and development teams who are building and deploying applications powered by large language models (LLMs).

Features

  • Warehouses LLM requests to a user-controlled PostgreSQL database for enhanced data ownership and security.
  • Offers a SQL interface to query logs for analyzing performance, troubleshooting issues, and calculating costs.
  • Caching capabilities to optimize costs and reduce latency on repetitive requests.
  • Supports one-time evaluations to test models against historical request logs.
  • Enables continuous monitoring of AI features in production with customizable alerts.
  • Facilitates the generation and export of datasets for fine-tuning and other training workflows.
  • Compatible with OpenAI and Anthropic endpoints, with support for other models on paid plans.
This profile is AI-generated and may contain inaccuracies.