On this page
Cohere: artificial intelligence features and setup
An enterprise AI platform with language models, information representations, and search result ranking. It is relevant to knowledge workflows that need to find suitable information before preparing an answer.
Tool by the Otomatzia editorial team · source content checked on
Overview
In a system that answers questions about organizational knowledge, a writing model does not replace a mechanism for finding the right document. Cohere's Embed and Rerank components are relevant to the retrieval and ranking steps before answer generation. Measure separately whether the right source was found and whether the answer used it correctly, so a search error is not mistaken for a language model writing problem.
Features
- Command models for language generation
- Embed for representing information for search
- Rerank for improving the order of retrieved results
Business uses
- Rank internal search results before showing them to an employee
- Prepare an answer from knowledge documents selected by a retrieval mechanism
Getting started
- Prepare knowledge questions with source documents and compare search results before and after ranking.
- Build a sample of questions with a known answer document and compare the order of results before and after Rerank, including questions in Hebrew that use a different term from the one in the document.
- Run a question from a user with limited permissions and ensure access filtering happens before sources are shown. Then check that the answer does not add a fact that the sources do not contain.
What to check
- Check Hebrew quality in the specific model and its suitability for your documents.
- Review deployment options and permissions to ensure retrieval does not expose a document the user is not allowed to read.
Sources to check next
Vendor features and prices change. Confirm details on the linked official pages before you rely on them.