Welcome back to our series on building the ultimate AI research assistant for our AI agent podcast! In our first post, we built a fantastic agent that could search the web to find the latest AI agent news for the agent factory podcast. But what if we want to add more specialized skills, like getting the real pulse from developer communities on Reddit? To do that, we need to upgrade our agent's design.
In this guide, we are going to level up our skills and refactor our simple agent into a powerful multi-agent system. We will build a "Manager" agent that directs a team of specialists, including one with a custom-built Reddit tool, to gather richer, more diverse insights.
By the end of this post, you'll have an even more powerful Trend Spotter agent that gets information from multiple sources. More importantly, you will learn the advanced skills needed to build complex agents with ADK. You will know how to: Build a scalable multi-agent system. Build a custom tool from any Python function (like our new Reddit tool). Create an orchestrator agent that delegates tasks to a team of specialists. Write advanced prompts to manage a multi-step, multi-tool workflow. Debug a multi-agent system using the ADK's powerful Trace view.
This architecture is the key to unlocking your agent's full potential. Let's get started!
To allow our agent to access Reddit programmatically, we need to get API credentials. This is free and only takes a minute. Navigate to Reddit Apps: Log in to your Reddit account and go to the app preferences page: . Create a New App: Scroll to the bottom and click the button that says "are you a developer? create an app…". Fill out the form: name: Trend Spotter Agent Select the script option for the application type. about url: You can leave this blank. redirect url: You must enter for this field. Click create app. You will now be taken to a new page showing your credentials. Set Environment Variables: For security, we'll store these credentials as environment variables. Your client ID is the string of characters right under "personal use script". Your client secret is the long string next to the secret label. Open your terminal and run the following export commands: Update Dependencies: Add the praw library to your requirements.txt file and install it. Install it from your terminal:
If you haven't done so already, in our previous blog we showed how to define the settings to tell ADK how to securely connect to your Google Cloud account to use services like Vertex AI and Google Search.
Run this one-time command. It will open a browser for you to sign in, allowing ADK to make authorized requests on your behalf.
To organize our new team of agents, we'll create a subagents directory inside our main trendspotter package.
No worries if you haven't gone through the first blog post, we got you! here is how you create the folder structure for your main agent:
After you have the main agent and your folder structure defined, from your trend-spotter root folder, run:
Now we'll build our two specialist agents by directly creating instances of the Agent class. The Google Search Specialist: Create a new file: trendspotter/subagents/googlesearchagent.py Add this code. Note how we directly create the googlesearchagent variable. The Reddit Specialist:
First, create a new file for our custom tool's code: trendspotter/tools.py. Add the following function to it.
Now we'll modify our main agent from Part 1 to become the "manager" of our new specialist team.
Open trendspotter/prompt.py and replace its contents with this new orchestrator prompt:
Open trendspotter/agent.py and replace its contents to turn it into the orchestrator. Notice how we now import the agent instances we created.
The run process remains the same. The adk web tool will automatically load your rootagent, which is now the orchestrator.
In the "Event" tab of the web UI, you will now see the main orchestrator agent making calls to its new googlesearchagent and redditagent sub-agents. You've successfully built a more robust and scalable AI system using the correct ADK patterns!
Congratulations! You have successfully upgraded your simple agent into a powerful, multi-agent system using the Agent Development Kit's orchestrator pattern. This is a huge step in your journey as an agent developer.