Vectara

Vectara is a modern, API-first search platform tailored for application developers and data engineers. It offers a Retrieval Augmented Generation-as-a-Service (RAGaaS) platform that includes breakthrough relevance, language-agnostic search, and secure, reliable search-as-a-service. Vectara supports cross-language search, allowing users to search in one language for content written in another language. The platform is designed to mitigate hallucinations, provide explainability, enforce access control, and allow for real-time updates of knowledge. It automatically extracts text from various formats such as PDF, Microsoft Word, HTML, and JSON, using zero-shot models optimized for language understanding to encode data at scale. Vectara segments data into indexes storing vector encodings optimized for low latency and high recall, and can recall candidate results from millions of documents using zero-shot neural network models. The platform increases the precision of retrieved results with cross-attentional neural networks to merge and reorder results. Vectara offers a free Growth plan with up to 15,000 queries and 50 MB account size per month, and a Scale plan that includes advanced RAG controls, custom dimensions, cross-language search, and premium support. The platform provides a factual consistency score with every response and supports indexing and searching in the world’s most commonly written 11 languages. Deployed as a fully managed cloud service on AWS, Vectara eliminates the need for server maintenance and upgrades, and can auto-scale to handle higher query volumes without additional infrastructure management. The platform includes features like snippet extraction, metadata annotation, and neural re-ranking, and supports encryption at rest and in transit, with options for customer-managed encryption keys. Vectara provides a complete GenAI experience from extraction, encoding, and indexing to retrieval and rerank, and offers a variety of sample applications and demos to showcase the power of generative AI in different use cases.