Automate Competitor Price Tracking with a Browser Agent

Keeping tabs on competitor pricing is crucial, but manually checking websites and updating spreadsheets is tedious and time-consuming. What if you could delegate this task to an AI agent that acts like a junior analyst?

In this tutorial, we'll walk through how to build exactly that using Filtyr. We'll create an AI agent that can:

1. Read a list of competitors and their pricing page URLs from a Google Sheet.
2. Visit each pricing page using a web browser.
3. Summarize the pricing information found.
4. Update the Google Sheet with the summary.

Let's get started!

Prerequisites:

• A Filtyr account.
• A Google Sheet set up with columns for Competitor, Pricing URL, and Pricing summary.
• Ensure the Google Sheet is shared with the email address your AI agent will use (we'll cover giving the agent access later).

(Example Google Sheet Setup)

Step 1: Create Your New Agent in Filtyr

1. Log in to your Filtyr dashboard.
2. Navigate to the Agents section in the left-hand sidebar.
3. Click the + New agent button (usually in the top-right corner).
4. A pop-up will appear. Give your agent a descriptive name, like Price tracking analyst.
5. (Optional) You can add a description, but we'll skip it for this example.
6. Click Create agent.

You'll be taken to the configuration page for your newly created agent.

Step 2: Give Your Agent Browser Access (with Authentication!)

For the agent to access websites and your private Google Sheet, it needs a web browser tool that's already logged into the correct Google account.

1. On your agent's configuration page, click the Browsers tab.
2. Under "Connected Browsers", click the + Connect Browser button.
3. You'll see a list of available "Browser Contexts". These are pre-configured browser sessions, often saved with specific logins. Select a browser context that is logged into the Google account that has edit access to your competitor tracking Google Sheet. (In the video example, this was named "Agent Gmail"). Note: You might need to set up a Browser Context separately in Filtyr if you haven't already.
4. Click the + Connect button next to your chosen browser context.

This browser is the primary tool our agent needs for this task. It doesn't require complex custom tools, just the ability to browse the web as an authenticated user.

Step 3: Instruct Your Agent

This is where the magic happens. Filtyr allows you to instruct agents using natural language, much like talking to a human assistant.

1. Go back to the Instructions tab for your agent.
2. In the "Agent Instructions" text box, we'll provide clear directions. Here’s the prompt used in the video:
   
   You are a competitor pricing junior analyst. You are tasked with tracking pricing of our competitors.

This google sheet has the list of our competitors, their pricing page URLs, and a space for a pricing summary (thats for you to fill out).

[Paste the Share Link to your Google Sheet Here - Ensure 'Editor' access for the agent's email]

# Detail the Steps
Here are loose steps you can follow:

1. Go to the sheet to get the first pricing url
2. Go to that url and get a summary the pricing data. Don't worry about the details. We just need the basic summary.
3. Update the pricing summary with a summary of that competitors pricing.

Do that for each competitor in the list.
   
3. Click Save.

Step 4: Define Outcomes (for Monitoring)

Outcomes help you categorize the results of the agent's work, making it easier to monitor success or failure.

1. Scroll down to the "Outcomes" section on the Instructions tab.
2. Click the + New button.
3. Create a Success outcome:
4. • Status: Enter Success
   • Color: Choose a visual indicator (like green).
   • (Optional) Describe when to use it.
   • Click Create decision.
4. Click + New again.
5. Create a Failed outcome:
6. • Status: Enter Failed
   • Color: Choose a different visual indicator (like red).
   • (Optional) Describe when to use it.
   • Click Create decision.

Step 5: Activate and Trigger Your Agent

Now it's time to set the agent loose!

1. At the top right of the agent configuration page, toggle the switch from "Inactive" to Active.
2. Navigate to the Triggers tab.
3. You have options:
4. • Schedule: This is ideal for production. You can set the agent to run automatically at specific times (e.g., "Weekly on Monday at midnight", "Monthly (1st of month)"). This automates the process completely.
   • Manual: This is perfect for testing. It allows you to run the agent immediately on demand.
4. For testing, click the Manual sub-tab.
5. Click the Run Agent button.

Step 6: Watch Your Agent Work (and Verify)

Filtyr provides a live workspace where you can watch the agent execute its task in real-time.

1. After triggering manually, you'll be taken to the agent workspace view.
2. On the right, you'll see the Live Browser View, showing exactly what the agent sees and does in the connected browser. You'll see it open the Google Sheet, navigate to competitor URLs, and scroll through pages.
3. On the left, the Agent Execution Trace shows the agent's thought process and the steps it's taking ("Use Tool: Agent Gmail Browser", "Now I'll update the pricing summary...", etc.).
4. Let the agent run through all the competitors listed in your sheet. This might take a few minutes.
5. Once finished, check your Google Sheet! The "Pricing summary" column should now be populated with the information gathered by your AI agent. You can see the agent's Google account ("Agent Filtyr" in the video) as the last editor.

Conclusion

Congratulations! You've successfully built and tested an AI agent to automate competitor price tracking. By simply providing natural language instructions and connecting the right tool (an authenticated browser), you've created a digital assistant to handle a repetitive task.

Now, you can confidently set up a Schedule Trigger (like weekly or monthly) to keep your competitor pricing information automatically updated, freeing up your time for more strategic work.

This is just one example, imagine the other repetitive browser-based tasks you could automate using similar techniques!

Want to see it in action? Watch the original video tutorial here:
https://www.youtube.com/watch?v=4p_BHuYt_ko