Skip to main content

Lab 36: Designing an Enterprise Agent Strategy Challenge

Goal​

In this lab, you will design a strategy for a fictional company to adopt and govern a fleet of AI agents using Gemini Enterprise. This is a conceptual exercise to help you think about the challenges of managing agents at scale. There is no coding in this lab.

Prerequisites​

  • Note on Costs: Gemini Enterprise services can incur costs. Please be aware of the pricing for these services.

The Scenario​

You are the lead AI architect at a large retail company. The company wants to leverage AI agents across several departments: Sales, Marketing, and HR. Your task is to create a high-level plan for how you would use Gemini Enterprise to build, deploy, and govern these agents.

Step 1: Identify the Agents​

First, brainstorm the types of agents each department might need. Think about which ones you could build with the ADK, and which ones might be available as pre-built Google agents.

  • Sales Team:
    • Example Idea: An ADK-built "Lead Qualifier Agent" that connects to Salesforce via a data connector to score new leads.
  • Marketing Team:
    • Example Idea: Use the pre-built "Idea Generation Agent" for brainstorming new campaigns.
  • HR Team:
    • Example Idea: An ADK-built "Policy Assistant Agent" that connects to the company's SharePoint to answer employee questions about HR policies.

Your Task: Come up with at least one more agent idea for each department.

Step 2: Plan the Data Connectors​

For your agents to be useful, they need access to the company's data.

Your Task: Based on the agents you identified above, list the Gemini Enterprise data connectors you would need to configure. For each connector, specify which agent would use it.

  • Example: Salesforce Connector for the Lead Qualifier Agent.

Step 3: Design the Governance and Access Control​

Not everyone should have access to every agent or every piece of data.

Your Task: Define a simple Role-Based Access Control (RBAC) policy.

  • Who should be able to use the Lead Qualifier Agent?
  • Who should be able to edit or manage the Policy Assistant Agent?
  • Should the Marketing Team be able to see data from the Salesforce Connector?

Step 4: Plan for Monitoring and Cost Management​

Finally, think about how you will monitor the system.

Your Task:

  • What is one key metric you would want to track for the Lead Qualifier Agent? (e.g., number of leads qualified per day).
  • What kind of alert would you set up for the Policy Assistant Agent? (e.g., alert if the agent fails to answer a question more than 10% of the time).
  • How would you set a budget for the Marketing team's usage of the "Idea Generation Agent"?

Lab Summary​

You have successfully created a high-level strategic plan for deploying and managing an enterprise-wide agent ecosystem.

You have learned to think about:

  • Identifying opportunities for specialized agents across a business.
  • The importance of data connectors for grounding agents in enterprise reality.
  • Designing governance and access control policies to ensure security.
  • Planning for monitoring, alerting, and cost management in a production environment.

Check the lab-solution.md to see example answers for this exercise.

Self-Reflection Questions​

  • Why is a centralized "Agent Gallery" a valuable feature for a large enterprise? What problems does it solve?
  • The "Agent Designer" is a no-code tool. What are the benefits of empowering non-developers to build their own simple agents, and what are the potential risks you would need to manage?
  • How does grounding an agent in a private, enterprise data source (like SharePoint) using a secure data connector mitigate the risk of the agent hallucinating or providing incorrect information?

🕵️ Hidden Solution 🕵️​

Looking for the solution? Here's a hint (Base64 decode me): L2RvYy1hZGstdHJhaW5pbmcvbW9kdWxlMzYtZ2VtaW5pLWVudGVycHJpc2UvbGFiLXNvbHV0aW9u

The direct link is: Lab Solution