Google revealed its Enterprise Search in Google I/O 23.
Enterprise Search acts as an AI search engineer for a company’s own data (as opposed to data indexed on Google and other search engines). It can be used as a Cloud service to build custom products (API access) or as an embeddable widget. Enterprise Search supports three types of use cases: a search engineer for a website’s content and structured or unstructured data. Under the hood, it slices custom data and documents and creates embeddings. As it uses LLM to surface information, all the usual functionalities of an LLM can be expected: Q&A, summarisation, recommendations and so on.
Google emphasised one interesting point as the differentiation factor to regular LLM chatbots - the search results will be strictly from the documents, and there will be no made-up responses.