A Copilot provides a conversational experience for users to ask questions and complete tasks within an application. Its behavior depends on its instructions, connected knowledge, and configured tools.
With AI-assisted creation, you can explain the Copilot’s purpose and responsibilities, then review the proposed configuration before creating it.
This article uses a procurement scenario with an existing Procurement Knowledge Knowledge Source and Procurement Inquiry Agent.
Before You Start
Open the application where you want to create the Copilot. Make sure you have permission to add application resources.
Before starting, identify:
Who will use the Copilot.
The questions and tasks it should support.
The knowledge and application resources it needs.
Any existing AI Agents it should use.
Which actions it is allowed to perform.
If you plan to reuse Procurement Knowledge or Procurement Inquiry Agent, check that they are configured and available for the intended use.
Step 1: Open New Application Resource
In your application, click Add resource (+) to open New application resource.
In the Create with AI area, select Copilot. Click Show more if the option is not immediately visible.
Step 2: Describe Your Copilot Requirements
Enter a description of the Copilot you want to create.
Explain its intended users, responsibilities, information sources, and action boundaries.
Replace the example names with resources available in your application.
For better results, include the following details:
Detail | What to include |
Copilot name | A clear name users will recognize |
Intended users | The team or audience it will support |
Responsibilities | Questions and tasks it should handle |
Knowledge | Existing Knowledge Sources it should use |
AI Agents | Existing Agents it should reuse, if needed |
Tools and actions | The capabilities required for its tasks |
Response style | Preferred language, tone, and level of detail |
Boundaries | Actions it should avoid and when to ask for clarification |
Click Start with AI to submit your request.
Step 3: Clarify Resources and Behavior
AI may ask questions before preparing the implementation plan.
For example, it may need to clarify:
Which Knowledge Source to use.
Whether an existing Agent should handle a task.
Which application resources are relevant.
Whether users need information retrieval or data-changing actions.
How the Copilot should respond when information is unavailable.
Answer using the actual resource names and business requirements.
Resolve unclear or unavailable dependencies before confirming creation.
Step 4: Review the Implementation Plan
AI prepares an Implementation Plan based on your requirements and any clarifications.
Review the proposed Copilot configuration carefully.
Review area | What to check |
Identity | The name and description explain the Copilot’s purpose |
Instructions | Responsibilities, response style, and boundaries match your needs |
Knowledge | The correct Knowledge Sources are proposed |
Agent connections | Any proposed Agent references point to the intended Agents |
Tools | Each tool is necessary for the requested tasks |
Access settings | Resource access and execution settings are appropriate |
Remaining setup | Any configuration still required is identified |
For the inquiry-only example, verify that the proposal does not include capabilities for modifying records or starting workflows.
If changes are needed, explain them in the conversation.
Review the revised plan before proceeding.
The Copilot is created only after you confirm the Implementation Plan.
Step 5: Confirm and Create the Copilot
When the plan meets your requirements, confirm it in the AI conversation.
AI creates the Copilot based on the confirmed plan. Follow any additional setup prompts shown in the interface.
When creation is complete, open the Copilot using the resource link or entry provided.
Step 6: Review the Copilot Configuration
Open the Copilot and check the generated settings.
Instructions
Confirm that the instructions clearly define:
The Copilot’s purpose and intended users.
The questions and tasks it should support.
How to use its knowledge, Agents, and tools.
How to handle missing or ambiguous information.
Its response style and action boundaries.
Knowledge
Check that the intended Knowledge Sources are connected and that the relevant content is available.
Tools and AI Agents
Review the configured tools and any connections to existing Agents.
Check target resources, parameters, and execution credentials where applicable. If the Copilot calls an Agent, review that Agent’s capabilities as well.
Instructions alone do not determine which actions are available. For an inquiry-only Copilot, verify that its tools and connected Agents are configured for the intended retrieval tasks.
Step 7: Test the Copilot
Use the available preview or testing interface to try representative questions.
Replace the placeholders with identifiers that exist in your application.
Check that:
Answers accurately reflect the available source information.
The Copilot selects appropriate knowledge, tools, or Agents.
Missing or ambiguous identifiers trigger clarification.
Unavailable information is clearly identified.
Source records are identified where possible.
The Copilot stays within its configured action boundaries.
Also test a request outside its intended scope to verify how it responds.
Step 8: Publish and Make the Copilot Available
After reviewing the configuration and completing testing, publish the Copilot using the available publishing controls.
Review its application installation, availability, and user access settings. Complete any additional setup required to make it available to the intended users.
Publishing and making a Copilot available in an application are related configuration steps. Verify both before announcing it to your team.
Continue Refining with Ask AI
After creation, use Ask AI to return to the original saved conversation and request further changes.
Review the resulting changes, test the affected scenarios, and publish the updated Copilot when ready.