From warehouse to AI-powered insights in minutes

Census, a Fivetran company, is expanding its capabilities to integrate artificial intelligence into data workflows by introducing two new features: Embedding Columns and new vector store connections. These innovations allow data teams to generate vector embeddings directly within datasets using models like OpenAI or Gemini, and sync them to vector databases such as Pinecone and turbopuffer, facilitating AI-ready data without the need for custom pipelines. This advancement addresses the complexity of preparing data for AI, streamlining the process by embedding it into existing Census workflows, making it easier to transform structured data into vector embeddings for use with large language models and retrieval systems. An example of this is enhancing support ticket searches, where embeddings enable the identification of semantically similar issues despite differences in phrasing, thereby improving issue resolution and support automation. These new capabilities leverage existing infrastructure, ensuring data freshness and flexibility, and positioning the data warehouse as a foundation for AI-driven applications, including CRM intelligence and intelligent support systems.