In 5 minutes, I'll show you how to build AI agents from scratch

Source: Dev.to

What Is an AI Agent? (Simple Explanation)

A normal AI model (like ChatGPT or Gemini) only gives you text answers.

An AI agent can:

• Perform tasks, not just answer questions
• Use tools, such as Google Search, calculators, APIs, or databases
• Make decisions and plan steps
• Work autonomously with minimal human help

Think of an AI agent as a small “AI worker” that can do tasks for you.

What You Need Before Starting

• Python installed on your computer
• Any code editor (e.g., VS Code or Google Antigravity IDE)
• Google ADK – the open‑source framework we will install (see the Google ADK overview)
• A Gemini API key (free to generate)

Build AI Agents From Scratch

Step 1: Create a Project Folder and Virtual Environment

mkdir my_first_agent
cd my_first_agent

Create a virtual environment:

python -m venv .venv

Activate it:

# macOS / Linux
source .venv/bin/activate  

# Windows
.\venv\Scripts\activate

A virtual environment keeps your project clean and avoids version conflicts.

Step 2: Install Google ADK

pip install google-adk

Step 3: Create Your Agent Project

adk create my_agent

Choose the model (e.g., Gemini 2.5 or Gemini Flash).

After creation you’ll see a folder structure:

my_agent/
 ├── agent.py
 ├── .env
 └── __init__.py

• agent.py – main file where your code lives
• .env – store your API key

Open the folder in your IDE.

Step 4: Get Your Free Gemini API Key

1. Go to and sign in with your Google account.
2. In the bottom‑left sidebar click Get API key.
3. Click Create API Key, give it a name, select (or create) a project, and copy the key.

Add the key to .env:

GOOGLE_API_KEY="your-key-here"

Step 5: Test the Default Agent

Open agent.py; you’ll see boilerplate code.
Replace the placeholder model name with the internal name of a Gemini model (e.g., gemini-2.0-flash or gemini-3-pro-preview).

Run the agent:

adk run my_agent

Ask a simple question. If everything works, you’ll receive an answer. If you hit the free limit for a model, switch to a lighter, free‑tier model.

Step 6: Create Multiple Agents (Research + Summarizer + Coordinator)

Add the following code to agent.py (or a new module) to build a multi‑agent pipeline:

from google.adk.agents.llm_agent import Agent
from google.adk.tools import google_search, AgentTool

# Research Agent – searches the web
research_agent = Agent(
    name="Researcher",
    model="gemini-2.5-flash-lite",
    instruction="""
You are a specialized research agent. Your only job is to use the
google_search tool to find the top 5 AI news items for a given topic.
Do not answer any user questions directly.
""",
    tools=[google_search],
    output_key="research_result",
)

print("Research Agent created successfully.")

# Summarizer Agent – creates summaries
summarizert = Agent(
    name="Summarizert",
    model="gemini-2.5-flash-lite",
    instruction="""
Read the research findings {research_result} and create a summary for each topic,
including a link to read more.
""",
    output_key="summary_result",
)

print("Summarizert Agent created successfully.")

# Root Coordinator – orchestrates the workflow
root_agent = Agent(
    model="gemini-2.5-flash-lite",
    name="root_agent",
    description="A helpful assistant for user questions.",
    instruction="""
You are the coordinator. First, delegate user questions to the 'research_agent' to gather information.
Second, pass the findings to the 'summarizert' agent to create a summary.
Finally, compile the summaries into a final response for the user.
""",
    tools=[
        AgentTool(research_agent),
        AgentTool(summarizert),
    ],
)

This defines a three‑step pipeline: research → summarization → coordination.

Step 7: Run the Multi‑Agent System

adk run my_agent

When prompted, enter the topic you want researched. You’ll see:

• The Research Agent gathering information
• The Summarizer Agent producing summaries
• The Coordinator compiling the final response

All agents work together to deliver the result.

Step 8: Use the Web Interface (Much Easier)

Google ADK includes a built‑in web UI that lets you interact with your agents through a browser, making testing and iteration even simpler. Launch it with the same adk run command and open the provided local URL in your browser.