PROFESSIONAL SERVICES

Executive AI Assistant

CHALLENGE

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.

CHALLENGE

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.

SOLUTION

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.”

SOLUTION

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.”

How it works

STEP 1.

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

STEP 2.

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)

STEP 3.

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

STEP 4.

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