Sellers at a mid-size distributor use the Copilot chat panel in Dynamics 365 Sales to ask questions like 'what were the last three interactions with this account.' During a pilot, several sellers report the chat sometimes returns interaction details that do not match anything in the account's actual activity history. A consultant is asked to reduce these incorrect answers before the pilot expands.
Select an answer to reveal the explanation.
Short Explanation
When an AI chat feature starts inventing details that were never logged anywhere, the fix usually is not to teach it more or lock it down further, it is to make sure it is actually looking at the real records before it answers. Grounding means tethering the AI's response to the account's genuine activity history so it reports what happened instead of guessing at a plausible-sounding pattern. The tempting wrong turns here all miss that distinction: retraining the core model is not something a consultant configures at the tenant level, so it does not even apply. Widening what data a seller is allowed to see changes visibility, not whether the AI is anchored to truth, so accuracy issues would persist. And simply switching the feature off avoids the embarrassment for now but leaves the same gap waiting the moment it comes back on. The real move is connecting the answers to the source of truth so every response traces back to something that actually happened on that account.
Full Explanation
The correct answer is C. When Copilot answers are not grounded in the organization's own Dataverse records, it can generate plausible-sounding but fabricated details instead of pulling actual logged activities, which explains why sellers are seeing interaction summaries that do not match reality. Configuring grounding so Copilot retrieves and cites the account's real activity records addresses the root cause directly. Option A is incorrect because tenant consultants and admins do not retrain the foundation large language model; that is not an available or appropriate configuration lever for this problem. Option B is incorrect because expanding security role privileges changes what data a seller can see, not whether Copilot's answers are anchored to real records versus generated text, so it would not fix fabricated content. Option D is incorrect because pausing the feature avoids the pilot's immediate risk but does not solve the underlying grounding gap, and the same inaccuracy would resurface whenever the feature is re-enabled without configuration changes. For a consultant, the key distinction is that answer accuracy in this scenario is a data-grounding configuration problem, not a model-training, permissions, or availability problem, and treating it as the wrong type of problem wastes rollout time without fixing the sellers' actual complaint.