A consultant is extending a custom Copilot Studio agent for a manufacturing distributor's Dynamics 365 Sales reps. Reps ask questions about detailed product specification sheets that are stored as PDF documents in a SharePoint document library and updated weekly by engineering — the specs are never entered into Dataverse. The consultant wants the agent to answer these questions accurately without manually authoring topics for every spec. What should the consultant configure?
Select an answer to reveal the explanation.
Short Explanation
Think about where the real, current information actually lives. The specs get updated every week by engineering and only exist as PDFs in a SharePoint library — they were never typed into the CRM. So the smartest move is to point the agent's generative answers straight at that library, letting it search and quote from the real documents whenever a rep asks. That way the agent is always as current as the source, with no extra maintenance. Copying everything into a new table just to search it there means building and babysitting a sync job that can drift out of date. Writing a dedicated script for every single spec question breaks down fast once you realize specs change constantly — you would be rewriting scripts every week. And routing this through the email-drafting assistant makes no sense, since that tool helps write emails, not answer open-ended product questions from a chatbot.
Full Explanation
The correct answer is D. Generative answers in Copilot Studio can be pointed directly at a SharePoint document library, letting the agent search and cite the PDF specs in place rather than requiring anyone to re-author or re-enter the content elsewhere. This matches the scenario: specs update weekly and live outside Dataverse, so the agent needs a knowledge source it can query live rather than a copy that must be kept in sync. Option A is incorrect because mirroring the library into a new Dataverse table adds a nightly sync process, introduces lag between engineering's updates and what reps see, and creates a second copy of the data to maintain for no functional gain over indexing the SharePoint source directly. Option B is incorrect because authoring a static topic per spec does not scale against a growing, weekly-changing set of documents and would require constant manual rewrites to stay current. Option C is incorrect because Copilot for Sales in Outlook is an email-drafting assistant, not the chatbot answering reps' spec questions in this scenario, so configuring it does nothing to solve the grounding problem the consultant is trying to fix.