DigitalOcean GenAI Platform lets you build GPU-powered AI agents with fully-managed deployment. Agents can use pre-built or custom foundation models, incorporate function and agent routes, and implement RAG pipelines with knowledge bases.
Guardrails are security components that you can attach to GenAI Platform agents. A guardrail overrides an agent’s output when it detects sensitive or inappropriate content in the agent’s input or output. For example, it can prevent an agent from sharing login credentials or credit card information. For more details, see our list of supported guardrails.
GenAI supports three guardrails: sensitive data, jailbreak, and content moderation. You can further customize these guardrails by duplicating and editing them.
You can attach guardrails to agents in two ways. To simultaneously attach multiple guardrails to the same agent, use the agent’s Resource page. To simultaneously attach multiple agents to the same guardrail, use the Guardrails page.
To simultaneously attach multiple guardrails to the same agent, go to the DigitalOcean Control Panel. In the left menu, click GenAI Platform, then click the Agents tab to view your agents page. In the Agents section, find the agent you want to attach guardrails to, then click the Resources tab. Scroll to the Guardrails section and click Add guardrails.
This opens the Add guardrails window. Check the boxes next to the guardrails you want to attach, then click Save to confirm your changes.
To simultaneously attach a guardrail to one or more agents, go to the DigitalOcean Control Panel. In the left menu, click GenAI Platform, then click Guardrails tab to view your guardrails page. To the right of the guardrail you want to attach, open the … menu, then click Attach Agent.
This opens the Manage Agents window. Check the boxes next to the agents you want to attach, then click Update to confirm your changes.