Autonomous AI agents come with risks, such as security threats, data breaches, reputational harm, financial loss, bias, hallucinations, and issues with transparency and accountability. When planning an AI agent, identify risks for your Agentforce use case and plan risk mitigation strategies.
Required Editions
| Available in: Lightning Experience |
| Available in: Enterprise, Performance, Unlimited, and Developer Editions. Required add-on licenses vary by agent type. |
As you approach conversations in your organization about AI risk, you can use the People, Business, Technology, and Data framework. These categories and considerations can help you come up with possible risks and concerns related to your Agentforce project.
| Category | Considerations |
|---|---|
| People |
|
| Business |
|
| Technology |
|
| Data |
|
Use this framework to discuss concerns related to your Agentforce use case. Your project stakeholders can brainstorm potential risks for each category.
After cataloging the risks and concerns, come up with mitigation strategies for each risk. Some of the mitigation strategies will translate into agent guardrails in Agentforce. When you’re done with the risk mitigation exercise, document the risks and guardrails for your use case. Capturing risk mitigation activities is important for regulatory compliance and useful for internal audits.
To see an example of risks and guardrails for an Agentforce project, complete the Define the Agent Guardrails unit on Trailhead.