Cohere, a Toronto-based startup founded by former Google employees, has entered the competitive AI chatbot market with the release of its new application programming interface (API). The API allows third-party developers to build chat applications using Cohere’s proprietary large language model (LLM), Command. The company aims to simplify the creation of reliable conversational AI products for enterprises. Cohere also provides its own free chatbot demo, Coral Showcase, for users to test out its chatbot, although access requires signing in with Google or Cohere credentials. In tests conducted by VentureBeat, the Coral chatbot powered by Command was slower than some competing chatbots but produced accurate and up-to-date responses without significant errors.
Cohere’s new chatbot API features Retrieval-Augmented Generation (RAG), which allows developers to control a chatbot’s information sources. Developers can constrain the sources to their enterprise data or expand them to include the entire internet while leveraging the chatbot’s training and text generation capabilities. Currently, Cohere supports two additional sources of information: web search implementation and plain text documents from the developer’s enterprise or other sources. The company plans to expand the chatbot’s capabilities and ecosystem by introducing features such as document mode, query-generation mode, and connector mode. Despite some limitations in the initial tests conducted by VentureBeat, Cohere aims to deliver high performance with its model trained specifically for RAG tasks.
Cohere’s entry into the AI chatbot market positions the startup as a leading player in the generative AI space. With its focus on developing foundation models and other AI-powered technologies for enterprises, Cohere aims to simplify the creation of chat applications through its new API. While its Coral chatbot powered by Command may be slower than some competitors, it offers accurate and up-to-date responses without visible errors. Cohere’s incorporation of Retrieval-Augmented Generation (RAG) allows developers to control and expand a chatbot’s information sources, enhancing the relevance and accuracy of generative AI responses. With plans to expand its capabilities further, Cohere is poised to make a significant impact in the AI chatbot market.