A small sample of the many use cases we are working on with customers, spanning nearly all industry sectors.

Use Cases | Cohere

Use Cases

Using Cohere models with retrieval-augmented generation (RAG), a luxury retailer built a virtual shopping assistant.

Cohere’s retrieval prioritizes accurate responses and citations

Cohere’s models come with connectors to common data sources

Cohere’s models can be fine-tuned to further improve domain performance

Cohere’s powerful inference frameworks optimize throughput and reduce compute requirements

Cohere models can be accessed through a SaaS API, on cloud infrastructure (Amazon SageMaker, Amazon Bedrock, OCI Data Science, Google Vertex AI, Azure AI), and private deployments (virtual private cloud and on-premises)

Over 100 languages are supported, so the same topics, products, and issues are identified the same way

Based on user interaction history, Cohere Command generates a query to search for content on the company platform

Connectors plug into customer’s  user profile database and content platform

Cohere Rerank sorts collected responses based on semantic relevance to the original query

Based on relevant responses, Cohere Command generates conversational wellness advice

Digital Health & Wellness AI Coach

A digital health company wanted to build a fully conversational health and wellness AI coach that considered user demographics, health goals, and content history. The goal was to increase user engagement and retention using the thousands of pieces of content they already had. They engaged Cohere to help them develop the coach application.

 with retrieval-augmented generation (RAG), the customer built an AI coach. By connecting to the company's content platform, the AI coach could draw from the company’s library of wellness advice as a source of truth, using this for conversations with users. A proof-of-concept was hosted via the Cohere Platform. The final production model will be hosted on a virtual private cloud service for the benefit of customer security and privacy.

Using retrieval-augmented generation (RAG) with Cohere models, a professional services firm built an AI assistant.

Internal AI Knowledge Assistant - Cohere Use Cases

Unstructured knowledge base documents are embedded by Embed and stored in a vector database.

Command interprets a user’s requests and creates queries across legacy and vector search sources.

Rerank re-orders the responses based on relevance to original queries, improving the accuracy of the search results.

Command synthesizes a conversational response, along with citations, back to the user.

Internal AI Knowledge Assistant

A large professional services firm wanted to build a knowledge assistant with retrieval-augmented generation (RAG)  using its knowledge base of tens of millions of documents, initially for internal users, and ultimately as a product for its clients. Its proof-of-concept had a combination of commercial generative models, open-source embeddings, and reranking solutions for search, but long-term, the customer was looking for a solution that could work in secure customer environments.

The customer replaced its open-source retrieval models with Cohere Embed and Cohere Rerank after testing them side-by-side and seeing substantial increases in retrieval accuracy. The team is currently replacing their generative model with a fine-tuned Cohere Command , making it possible to deploy this solution anywhere within their own network and on any client environment while staying fully compliant with demanding security requirements.

Using Cohere Rerank to improve search and retrieval for an AI assistant with retrieval-augment generation (RAG).

Cohere’s embedding prioritizes accurate reranking, even with noisy datasets

Rerank comes with connectors to common data sources

Rerank can be fine-tuned to further improve domain performance

An upstream generative model generates search terms based on user requests and conversation context

The customer’s knowledge base is queried using legacy search tools

Cohere Rerank sorts results based on semantic relevance to the original query

A downstream generative model uses the most relevant results to generate a conversational answer

Improved Search for AI Assistant

A major provider of digital workforce collaboration and productivity tools found their existing search tools did not perform well across customer-generated knowledge bases and the large volume of new customer-generated content meant embedding was not viable. The company decided it wanted to improve their retrieval-augmented generation solution with better search results.

 was integrated with the customer’s search systems. Rerank used output from existing search tools and reordered the results for use by the company’s Q&A app to provide relevant and more accurate answers.

Using Cohere models with retrieval-augmented generation (RAG), an asset management company built an AI assistant. 

Less time spent finding information by analysts

Faster research and due diligence process

AI Knowledge Assistant for Asset Management - Cohere Use Cases

Cohere Command generates queries for the cloud repository based on questions and conversation history

Cohere Rerank ranks repository search results by their relevance to user questions

Cohere Command generates a conversational answer, along with citations to source documents

AI Knowledge Assistant for Asset Management

A leading North American asset management company found its analysts were spending a lot of time painstakingly sifting through news, commentary, and financial reports from their cloud-based repository to give the right information in response to customer questions about company performance.

 with retrieval- augmented generation (RAG) to build an AI knowledge assistant connected to its cloud repository. Analysts could immediately ask the assistant questions in plain English and receive conversational responses they could use to answer customer questions, along with citations to the source documents within the repository.

Using Cohere models with retrieval-augmented generation (RAG), a retail chain built an AI assistant to support HR. 

Cohere Command builds a query based on an employee’s level, union affiliation, and search terms for employment contracts

The database is queried to find the employee’s union affiliation, and then manuals and employment contracts are searched

Cohere Rerank sorts collected results based on semantic relevance to the original query

Based on the most relevant documents, Command generates a conversational response with citations

Employee Support AI Assistant

A large North American retail chain engaged Cohere because it wanted to support its managers and employees with quick answers to common human resources and operational questions. Some answers weren’t always obvious and busy managers would often generate tickets that would overwhelm IT and take weeks to answer.

 with retrieval-augmented generation (RAG), the customer built a search app connected to the company’s HR database. Users could ask questions in plain language and receive simple, conversational responses based on location, union affiliation, and job level.

Using Cohere Rerank, a VC funded company optimized their AI knowledge assistant with improved search and retrieval.

Cohere’s models can be fine-tuned to improve domain performance further

A generative model creates queries for connected data sources based on user requests and conversational history

Cohere Rerank model sorts search responses by relevance to the original query along with relevance scores

A generative model uses Rerank responses to generate answers

Improved Search for AI Knowledge Assistant

A VC funded company needed to improve the performance of their AI knowledge assistant. It used the app to research and extract information from reports, financial statements, and other internal documentation. The company had worked out that implementing a full semantic search with embeddings was too costly and it would introduce some latency to the solution.

 to their existing search solution. It immediately provided better and more relevant results and improved the downstream performance of the generative model.

Using a fine-tuned Command, a video conferencing customer added AI generated summaries and action items to calls.

Higher end-customer satisfaction and retention

Meeting Summaries and Action Items - Cohere Use Cases

The fine-tuned Command model ingests a meeting transcript

Command delivers a separate list of action items with owner names

Meeting Summaries and Action Items

A global provider of video conferencing services needed to add call summarization and action item extraction features to its video conferencing offering in response to strong customer demand. Previous attempts to use generative AI with competitive models had been difficult to customize, and they were frequently inaccurate and too verbose for customer requirements.

The customer built a solution by fine-tuning Cohere’s Command model on the nuances of transcript summarization. It accurately summarized live meetings in bullet points, including action items for call participants, and delivered it on the company’s proprietary conferencing platform. Command delivered higher quality output than competitor models out of the box, and within two weeks of fine-tuning, achieved 4:1 preference on output quality.

Using Cohere Rerank, a SaaS business optimized their legacy search system for more relevant findings.

Existing lexical and semantic search tools are queried with the user search term

Cohere Rerank embeds the original query and search results in real time

Rerank returns matches re-ordered by semantic relevance to the original query, with relevance scores

Improved Search

A major provider of SaaS workforce collaboration and productivity tools wanted to improve its search capabilities. Its customers were often frustrated trying to find documents with its legacy search tools.

 was integrated into the company’s search infrastructure. Rerank ingested the output of the existing search tools, reordering the responses based on relevance to the user’s search query, resulting in far more relevant answers and higher customer satisfaction.

Using retrieval-augmented generation (RAG) with Cohere models, an insurance company built an underwriting AI assistant.

Ability to underwrite faster and increase business

Better use of highly skilled employees’ time

Insurance Underwriting Assistant - Cohere Use Cases

Insurance Underwriting Assistant

A large, multinational, commercial insurance company wanted to improve underwriting efficiency. Its skilled staff were spending many days reviewing engineering reports and other materials, and the company wanted to build a generative AI solution to identify and extract risk factors and highlights, help staff make faster and more accurate decisions, and hence speed up the underwriting process.

Initially, the company built a solution to summarize and extract relevant information from engineering reports with a  fine-tuned Cohere Command model hosted on Amazon Bedrock. In the second phase, the team is developing a full retrieval-augmented generation (RAG) solution using Cohere Embed and a vector database to embed engineering reports. The result will be an AI assistant that makes it easier for users to extract critical information by using a natural language interface and semantic search.

Better project staffing and client satisfaction

Knowledge and Project Staffing AI Assistant - Cohere Use Cases

Knowledge and Project Staffing AI Assistant

A global consultancy wanted to build an internal knowledge assistant using retrieval-augmented generation (RAG) to help its consultants find information and generate reports from millions of proprietary documents. The firm also wanted the assistant to help optimize project staffing with an intelligent directory tool that can quickly find and provide details on consultants with specific domain knowledge.

Cohere worked with the global consultancy to build a custom RAG solution based on Cohere Command, Embed, and Rerank. Using an intelligent assistant tool powered by Cohere models with RAG, consultants can ask conversational questions and get fast, accurate answers, along with citations to previous work by the experts – all based on millions of unstructured documents, notes, and authorship data.

Using a fine-tuned Command, a procurement firm created a tool to generate AI summaries from multi-modal documents.

Procurement Summaries - Cohere Use Cases

The fine-tuned Command model ingests unstructured notes, presentations, and transcripts.

Command produces a templatized summary of procurement considerations.

Procurement Summaries

A procurement consulting firm wanted to build a tool to summarize historical vendor procurement documents to help in future procurement discussions and optimize contract terms. The most valuable information was a supplier’s priorities and historical performance, but this information was often spread across different presentation decks, notes, and documents. Extracting all of this information was time-consuming for staff and required expertise.

The customer is currently working with Cohere to fine-tune a Command model that can ingest and summarize multi-modal files, such as PowerPoint decks, notes, and PDF documents. The goal is to train the model using previous contract negotiation materials and previous summaries compiled by skilled procurement staff.

Using retrieval-augmented generation (RAG) with Cohere models, a network security company built an AI assistant.

Less effort required by users to find answers

Accurate answers to domain-specific questions

Security Documentation AI Assistant - Cohere Use Cases

Unstructured policy documents are embedded by Embed and stored in a vector database.

Command interprets a user’s requests and creates queries to retrieve relevant answers.

Rerank re-orders the retrieved answers based on relevance to original queries, improving the accuracy of the search results.

Command synthesizes a conversational response back to the user, incorporating answers along with citations.

Security Documentation AI Assistant

A global network security company wanted to build an AI assistant to help its IT professionals respond to queries on network security setup and security best practices. Its security policy documentation repository includes tens of thousands of articles consisting of unstructured data, and its legacy search tools often produced inaccurate results. The company initially piloted a proof-of-concept (POC) using a competitor’s models, but their high latency meant that generated responses were too slow to meet user testing requirements.

The customer built the AI assistant with retrieval-augmented generation (RAG). Using Cohere Embed for semantic search to generate embeddings of the security policy documentation and Cohere Rerank to further improve retrieval accuracy by selecting the most relevant documents for answer generation. The team then fine-tuned Cohere Command to optimize how it answered user questions. The entire solution delivered the accuracy required at 50% lower latency than previous trials.

Using retrieval-augmented generation (RAG) with fine-tuned Cohere models, an investment platform built an AI assistant.

Investment Research Assistant - Cohere Use Cases

Unstructured financial documents are embedded by Embed and stored in a vector database.

Fine-tuned Command interprets a user’s requests and creates queries to retrieve relevant answers across legacy and vector search sources.

Fine-tuned Rerank re-orders responses based on relevance, improving the accuracy of search results.

Fine-tuned Command synthesizes a conversational response back to the user, incorporating answers along with citations.

Investment Research Assistant

A financial research platform for investment firms wanted to build a natural language interface for its clients to ask complex questions and get synthesized answers across financial reports, analyst research, investor call transcripts, and other data. The company initially needed to improve search relevance to boost client productivity and satisfaction, and then it wanted to build a full conversational AI assistant with retrieval-augmented generation (RAG).

The company discovered Cohere Rerank  through the AWS Marketplace. They fine-tuned Rerank and  implemented it, which provided an immediate improvement in search relevance on top of existing legacy search tools. Subsequently, the customer is implementing an embedding solution using Cohere Embed to further improve search performance. Lastly, they will build the AI assistant with a fine-tuned Command that will allow users to ask questions and get synthesized answers with citations in a natural language interface.

Using retrieval-augmented generation (RAG) with Cohere models, a CRM SaaS company built a tech support AI assistant.

Cohere Command generates queries based on conversation context and support questions

Connectors query support documentation site using legacy internal search tools

Based on relevant responses, Cohere Command generates a conversational response to questions, along with citations

Tech Support Assistant

A major provider of CRM and ERP software wanted to provide a better support Q&A experience that could report on transactional data. To achieve this, it worked with Cohere to build a support assistant that could answer common questions with conversational responses based on product documentation and internal knowledge bases.

 with retrieval-augmented generation (RAG) to build a conversational support app. Users could ask detailed technical questions and get relevant responses, along with citations.

Using Cohere Embed, retailers can build tools to quickly segment and analyze customer feedback.

Cohere’s retrieval performance ensures accurate responses and source citations.

Cohere models come with connectors to common data sources

Our models can be fine-tuned to further improve domain performance

 supports data compression, reducing storage and compute requirements

Cohere models can be accessed through a SaaS API, on cloud infrastructure (Amazon SageMaker, Amazon Bedrock, OCI Data Science, Google Cloud Platform, Azure AI), and private deployments (virtual private cloud and on-premises)

Cohere Embed processes input text and stores it in a vector database

Embeddings are clustered, classified, and named

Clusters are visualized, giving users content information, names, and trends across multiple languages

Customer Feedback Segmentation and Analysis

Multinational online retailers often struggle to act on product feedback. Products are often launched in multiple countries simultaneously, and synthesizing feedback across multiple languages can be difficult, time-consuming, and it’s easy to miss issues.

, retailers can build tools that can ingest text from support tickets, reviews, and third-party sources in over 100 languages. It can cluster by different attributes and name those clusters across languages giving retailers insights into feedback without extensive analysis or translation effort.

Virtual Shopping Assistant

A major luxury goods retailer wanted to help its client advisors, reducing the time advisors needed to find and research products. It engaged Cohere to create a tool that could help advisors suggest products to customers based on demographics, purchase history, and personal preferences.

Using retrieval-augmented generation (RAG), 

 with were integrated into the customer’s website, product catalog, inventory, and customer service apps to build a conversational virtual shopping assistant. The assistant can quickly give advice and relevant suggestions, helping advisors improve customer experience and basket size.

Using retrieval-augmented generation (RAG) with Cohere models, a consultancy built an AI executive assistant.

Immediate answers to complex, real-time questions

Executive AI Assistant - Cohere Use Cases

Unstructured knowledge base documents are embedded by Embed and stored in a vector database

Command creates a structured work plan and queries to answer a user’s request to retrieve information from connected data sources and tools (e.g., calculators)

Rerank re-orders the responses from search tools based on relevance to original queries, improving the accuracy of the search results

Command triggers another step in the work plan if required, otherwise it synthesizes a conversational response, along with citations, back to the user

Executive AI Assistant

A global financial consultancy wanted to build an executive AI assistant for a national telecom company that could support executive decision-making. To meet the client’s requirements, the solution needed to extract information from internal document stores and real-time data sources, and then perform calculations on the fly. Therefore, a purely model-based solution would be insufficient.

Using Cohere Command,  Embed, and Rerank, with  retrieval-augmented generation (RAG), the firm was able to  leverage Command’s multi-step capabilities and external agents (e.g., calculators and stock price sources) to retrieve and manipulate structured data. With this custom solution, the models break retrieval and computation tasks into multiple steps, allowing executives to ask complex requests like “Explain how my services margins compare to regional competitors.”