Skip to main content

Agent creation guidelines

When developing a new AI agent for your SRE platform using the Golem design system, follow these key guidelines to ensure effectiveness, ethical behavior, and user trust.

Define Clear Scope and Boundaries​

Clearly outline the agent's specific responsibilities and limitations. For example, a Monitoring Agent should focus on observing system metrics and alerting on anomalies, but not attempt to diagnose or fix issues.

Adhere to Ethical AI Principles​

Align the agent's behavior with Golem's core ethical principles:

  • Prioritize user well-being and privacy
  • Respect human agency and control
  • Ensure transparency in decision-making processes

Complement Human Expertise​

Design the agent to augment human capabilities, not replace them. Provide insights and suggestions that support SRE decision-making, while allowing humans to make final judgments on critical actions.

Implement Safeguards​

Build in checks and balances to prevent the agent from taking potentially harmful actions:

  • Require human approval for high-impact changes
  • Set strict limits on the agent's authority
  • Implement automatic rollback mechanisms for agent-initiated actions

Establish Feedback Mechanisms​

Create clear channels for users to provide feedback on the agent's performance:

  • Include rating systems for agent suggestions
  • Allow users to flag incorrect or unhelpful responses
  • Use this feedback to continuously improve the agent's capabilities

By following these guidelines, you'll create AI agents that are reliable, trustworthy, and truly beneficial to your SRE team.