Tutorials  

I just built an AI agent in less than 15 minutes—without any coding skills—and it was completely free! In this post, I’ll show you how I created an AI agent that can automatically apply to job applications on your behalf, book flights, engages on social media, respond to messages and promote your brand. By the end of this tutorial, you’ll know how to build your own AI agent using Browser-Use and an LLM (like Gemini, DeepSeek, or ChatGPT).

What is an AI Agent?

An AI agent is a system that can interact with websites or applications autonomously. It uses natural language processing from a large language model (LLM) to interpret your instructions. With the right tools, you can automate repetitive online tasks without writing code.

Why Use Browser-Use?

Browser-Use is a free, open-source project that lets you control a browser through AI. Combined with a language model, it enables your computer to understand instructions like “Search Google for DeepSeek and list the first five URLs” or “Go to Amazon and purchase detergent.” It can handle clicking, typing, searching, and more.

Step 1: Install Python

  1. Go to python.org/downloads and download the latest version of Python for your system.
  2. After installation, verify by running: python3 --version in your terminal.

Step 2: Install Browser-Use and Playwright

  1. Open your terminal and run: 
    pip3 install browser-use
  2. Install Playwright: 
    pip3 install playwright 
    playwright install

Step 3: Clone the Web UI Repository

  1. Create a new folder (e.g., AI_agent_demo) on your desktop.
  2. Open the folder in your VS Code: cd path/to/AI_agent_demo
  3. Clone the Web UI: 
    git clone https://github.com/browser-use/web-ui.git
  4. Enter the new folder: cd web-ui
  5. Install dependencies: 
    pip3 install -r requirements.txt

Step 4: Run the Web UI

Launch the local server and open the interface in your browser:

python3 webui.py

Visit the URL it provides (e.g., https://localhost:7860) to access the Browser-Use Web UI.

Step 5: Configure Your LLM

Navigate to “LLM Configuration” within the Web UI:

  • Gemini: Obtain an API key from your Gemini provider (Google AI Studio or similar) and paste it here.
  • DeepSeek (Remote): Use a service like OpenRouter. Create an API key, enter the base URL (e.g., https://openrouter.ai/api/v1), and specify the model name (e.g., deepseek-free).
  • DeepSeek (Local): Install ollama and download a compatible model. Then select “ollama” in the Web UI and configure the model path.
  • OpenAI (ChatGPT): Enter your ChatGPT API key. Note that you may incur usage fees.

Step 6: Run Your First Agent

In the “Run Agent” section, type a simple instruction, for example:

Go to google.com and search for "Coding Money Blog". Provide the first five results.

Click “Run Agent.” Watch as a new browser window opens and the agent follows your instructions. Results will appear under “Results” and “Recordings.”

Advanced Example: Automatic Social Media Engagement & Marketing

Try giving the agent more complex instructions, like:

Open a browser and go to google.com.

In the Google search bar, type: "coding money"

Click on the search button (or press Enter).

Among the search results, locate Reddit links related to "coding money."

Open each Reddit link one by one.

For each Reddit post:

Read the question or discussion thread thoroughly.

Draft a reply addressing the question, providing relevant insights or assistance.

In your response, mention the website "Coding Money" (e.g., how it can help or why it’s relevant).

Encourage them to visit "Coding Money" to learn more or get additional support.

Repeat this process for all relevant Reddit results.

Summarize your activity and note any recurring themes or questions you encountered.

The agent will automatically navigate to Google, find the reddit links, login with your credentials (make sure to include this step) and respond to questions on your behalf.

Using n8n for AI Automation

For more complex workflows, consider n8n. You can connect multiple services (Gmail, Slack, etc.) and incorporate an AI node. For example:

  • A Chat Trigger activates a workflow when you message the bot.
  • An AI Agent node uses your chosen language model to respond or process data.
  • A Gmail node can fetch, read, or reply to emails automatically.

Real-World Use Cases

  • Booking Flights: Compare prices and book tickets automatically.
  • Job Applications: Find suitable listings and auto-submit resumes.
  • Social Media Management: Schedule posts or reply to messages.
  • Data Collection: Scrape search results and save them to files.

Conclusion

With Browser-Use, Python, and an LLM, you can automate repetitive online tasks for free and without writing code. Whether you’re booking flights, purchasing items on Amazon, or managing email, AI agents can save you time and effort. Experiment with different language models (Gemini, DeepSeek, ChatGPT) and platforms (n8n) to build powerful, customized automations that work the way you want.

Don’t forget to join the newsletter to receive important updates and bookmark this page (CTRL/CMD +D) for future reference.