Deploy a self-contained agent to Gemini Enterprise Agent Platform to provide scalability, AI safety, and content moderation.
This guide helps you understand the Simple Agent Platform application template. You can use this template to quickly deploy your agent, along with safety settings, as an application on Google Cloud.
For example, you might implement this template to address the following business needs:
| Example | Business need | Implementation |
|---|---|---|
| Internal HR policy bot | An organization needs a secure HR assistant to answer employee questions while protecting sensitive data and maintaining professional language. | Sensitive data protection prevents PII leaks, while safety filters maintain professional tone. The serverless runtime scales to handle demand. |
| Secure coding assistant | A development firm needs an AI tool to explain code snippets while preventing jailbreak attempts and malicious prompt injection. | Model Armor blocks adversarial attacks and jailbreaks, while enforcement policies sever connections when malicious payloads are detected. |
| Multilingual public transit guide | A transit authority needs a multilingual assistant for tourists that strictly filters offensive content across all supported languages. | Multilanguage detection applies safety filters to all queries. A managed identity limits the agent's access to transit-related data. |
Architecture
The following image shows the template components and connections:
The template consists of the following components:
Agent Runtime: hosts the agent logic in a managed serverless environment. The environment can scale up to 3 instances to maintain performance. System performance is captured for monitoring and debugging.
Model Armor Template: inspects the model's input and output to block sensitive data leaks, jailbreak attempts, or toxic content. Use the
model_armor_template_MODEL_ARMOR_TEMPLATE_IDvariable in your agent code to apply safety policies to agent inputs and outputs.
What's next
- Duplicate and customize templates by building on Google templates.
- Define your own configurations by designing application templates.
- Plan your AI application strategy by choosing your agentic architecture components.