Guardrails AI is an open-source framework for validating and correcting the outputs of AI models, founded in 2023 by CEO Shreya Rajpal along with Diego Oppenheimer, Safeer Mohiuddin, and Zayd Simjee. Rajpal previously worked on machine learning at Apple and Drive.ai, while Oppenheimer founded data-platform company Algorithmia; the company raised $7.5 million in seed funding from Zetta Venture Partners, Bloomberg Beta, and Pear VC.
The framework adds a validation layer around AI inputs and outputs, checking generated text against reusable 'validators' that catch problems like hallucinated facts, jailbreak attempts, data leakage, and policy violations before they reach an end user. Guardrails Hub, its open-source library, now includes more than 100 community-contributed validators covering chatbots, RAG pipelines, summarization, and customer support use cases.
It suits engineering teams building production LLM applications who need output reliability controls without building a validation layer from scratch, with adoption reported at around 40,000 users including Robinhood. The core framework is free and open source under Apache 2.0; a commercial layer called Snowglobe adds hosted validation, observability, and enterprise support, with pricing available on request.
