Reference: Unsupported Custom AI Question Types
Contract Management and Document Intelligence includes security measures and design limitations to prevent the Custom AI model from answering certain types of questions. The model can't answer questions that require information outside the context of the provided documents or that fall into restricted categories.
This topic provides examples of unsupported question types and the model's expected response.
Question Type | Sample Question | Expected AI Response |
|---|---|---|
Factual (External) | What is the weather in San Francisco today? | I can’t provide real-time information such as weather forecasts. |
Factual (Tenant Data) | How many users are in this tenant? | I can’t access tenant-level administrative data, such as user counts. |
Counting | How many documents in the system are NDAs? | I can’t perform cross-document counting. To count documents, create a report with the same scope criteria used by the model. |
Aggregation | What's the total contract value of all active contracts expiring in the next 90 days? | I can’t aggregate data across multiple documents. To calculate totals, create a report that sources this data. |
Legal Advice | Can you provide a legal opinion on which contracts are risky? | I can’t provide legal advice or risk assessment. Consult with a legal professional. |
Proprietary | What large language model are you using? | What large language model are you using? I am a large language model developed by Workday. I cannot share details about my underlying architecture. |
Technical | Can you write Python code for me? | I can’t generate code in any programming language. |
Confidential | List all documents in folders that I can't access. | I can’t access or report on information that you do not have the security permissions to view. |