Toronto-based startup Cohere, founded by former Google employees, has entered the AI chatbot race by releasing a new application programming interface (API). This API allows third-party developers to build chat applications based on Cohere’s proprietary large language model (LLM), Command. The API, called Chat API, aims to simplify the creation of reliable conversational AI products for things like knowledge assistants and customer support systems. Cohere has also provided its own chatbot demo, called Coral Showcase, for users to test out the chatbot. In VentureBeat’s tests, the Coral chatbot powered by Command was slightly slower at generating responses compared to other closed-source chatbots, but the responses were accurate and up-to-date.
Cohere’s Chat API incorporates Retrieval-Augmented Generation (RAG), a method that allows developers to constrain chatbot information sources to their own enterprise data or expand them to include the entire web. Developers can add a web search implementation or plain text documents from their enterprise as additional information sources. Cohere plans to expand the capabilities of its connector/modular ecosystem in the future. VentureBeat’s limited initial tests found that the reliability of Cohere’s chatbot was not always as expected, but the tests were limited in scope and further testing is needed.
The release of Cohere’s Chat API comes shortly after OpenAI’s reintroduction of web browsing capabilities to its ChatGPT, as well as the announcement of its ChatGPT for Enterprise subscription service tier. Cohere aims to differentiate itself in the AI chatbot marketplace by focusing on developing foundation models and other AI-powered technologies for enterprises.