Automating Competitive Price Intelligence: Transforming Ecommerce with Amazon Nova Act

The Hidden Costs of Manual Competitive Price Intelligence

Automating with Amazon Nova Act

Common Building Blocks of Nova Act

Extracting Information from a Webpage

Navigating to a Webpage

Running Multiple Sessions in Parallel

Handling Captchas

Managing Errors

Building and Monitoring Nova Act Workflows

Building with AI-Powered IDEs

Observing Workflows in the Nova Act Console

Running the Solution

1. Prerequisites

2. Get Nova Act API Key

3. Clone the Repo and Set Up

4. Running the Script

5. Reviewing the Output

Conclusion

About the Authors

Automate Your Competitive Price Intelligence with Amazon Nova Act

Monitoring competitor prices is crucial for eCommerce teams looking to stay ahead in a competitive landscape. Yet, many teams are still bogged down by manual tracking—an inefficient method that consumes hours daily. This approach delays decision-making, inflates operational costs, and opens the door to human errors that could lead to lost revenue and missed opportunities.

This blog post explores how to leverage Amazon Nova Act, an open-source browser automation SDK, to create an automated competitive price intelligence system. By streamlining manual workflows, your team can make data-driven pricing decisions supported by real-time market insights.

The Hidden Cost of Manual Competitive Price Intelligence

For eCommerce teams, timely and accurate market data is essential. Legacy workflows often involve manually scouring various competitor websites to gather pricing information, recording it on spreadsheets, and consolidating that data for analysis. This method has multiple pitfalls:

1. Time and Resource Consumption

Manual price monitoring eats away hours each day, translating into significant operational costs that grow as your product catalog expands.

2. Data Quality Issues

Human error in data entry can lead to inconsistencies, potentially resulting in poor pricing decisions based on flawed data.

3. Scalability Limitations

As product ranges grow, relying on manual processes becomes increasingly untenable, creating bottlenecks in competitive analysis.

4. Delayed Insights

The most pressing issue is timing. Prices can fluctuate throughout the day, and stale data can lead to missed revenue opportunities.

These challenges are not exclusive to eCommerce; sectors like insurance, financial services, and travel face similar manual workflows that limit efficiency and competitiveness.

Automating with Amazon Nova Act

Amazon Nova Act is designed to empower developers to build intelligent agents that can automate web interactions using a natural language interface.

Key Features:

• Python-Driven Automations: Developers can create targeted commands using Python, allowing for enhanced programmatic logic alongside natural language instructions.
• Supporting Agentic Commerce Scenarios: Nova Act excels in automating tasks like competitive monitoring, content validation, and multi-step browsing workflows, making it particularly suitable for price intelligence.

Why Choose Nova Act?

Ecommerce websites frequently update their layouts and promotional banners, breaking traditional automation scripts. Nova Act’s innovative, flexible command structure allows agents to adapt to these changes seamlessly.

Common Building Blocks

1. Extracting Information from a Webpage

With Nova Act, agents can gather structured data directly from rendered pages. Define a schema using Pydantic models and utilize the act_get() call to extract data.

nova.act_get("Search for 'iPad Pro 13-inch (M4 chip), 256GB Wi-Fi'.", schema=ProductData.model_json_schema())

2. Navigating to a Webpage

Start a new session at a specific URL for data extraction.

nova.go_to_url(website_url)

3. Running Multiple Sessions in Parallel

Nova Act instances can operate concurrently, enabling concurrent checks across numerous competitor sites, significantly reducing monitoring time.

from concurrent.futures import ThreadPoolExecutor, as_completed

all_prices = []

with ThreadPoolExecutor(max_workers=10) as executor:
    future_to_source = {
        executor.submit(check_source_price, product_name, source_name, source_url, headless): source_name
        for source_name, source_url in sources
    }
    for future in as_completed(future_to_source):
        try:
            source = future_to_source[future]
            source_price = future.result()
            if source_price is not None:
                all_prices.extend(source_price.source)
        except ActError as exc:
            print(f"Skipping source price due to error: {exc}")

print(f"Found {len(all_prices)} source prices.")

4. Handling Captchas

Some websites may present CAPTCHA challenges during automated browsing. It’s ethically advisable to involve a human in these situations to solve such challenges.

5. Error Handling

Encounters with dynamic layouts or missing elements can result in errors during automation. Nova Act surfaces these as ActErrors, allowing developers to implement retry logic or fallback mechanisms.

Building and Monitoring Nova Act Workflows

Using AI-Powered IDEs

Accelerate your development process by leveraging AI-powered IDE extensions for Nova Act. These tools can turn natural language prompts into production-ready code, simplifying automation tasks.

Observing Workflows in the Nova Act Console

Monitor your automation flows in real-time via the Nova Act AWS console. With easy access to logs, screenshots, and performance metrics, this central management hub streamlines troubleshooting and accelerates the deployment process.

Running the Solution

To facilitate your journey into automated market research, we’ve released a Python sample project that tackles price monitoring.

Prerequisites

Ensure your development environment has:

• Python: 3.10 or later
• Nova Act SDK

Getting Started

1. Get the Nova Act API Key: Generate it from Nova Act Playground.

2. Clone the Repository:

   git clone https://github.com/aws-samples/amazon-nova-samples.git
   cd nova-act/usecases/price_comparison

3. Set the API Key:

   export NOVA_ACT_API_KEY="your_api_key"

4. Install Dependencies:

   pip install -r requirements.txt

Running the Script

Invoke the agent with the command:

python -m main.py --product_name "iPad Pro 13-inch, 256GB Wi-Fi" --headless

This command launches multiple Nova Act browser sessions, checks for the product across various sites, extracts prices, and consolidates the data for you.

Conclusion

Amazon Nova Act revolutionizes browser automation by transforming it into a user-friendly, natural language interface. Retailers can automate labor-intensive workflows, reduce operational costs, and gain immediate market insights.

By significantly cutting down the time spent on manual data collection, teams can direct their attention to strategic pricing decisions, contributing to a more competitive edge.

We would love to hear how you implement Nova Act in your workflows. Share your experiences in the comments or check out the Amazon Nova Samples GitHub repository for more examples.

About the Authors

• Nishant Dhiman is a Senior Solutions Architect at AWS based in Sydney, specializing in Serverless and Generative AI.

• Nicholas Moore is a Solutions Architect at AWS, focused on AI, analytics, and cloud solutions.

• Aman Sharma is an Enterprise Solutions Architect at AWS, dedicated to helping retail and supply chain customers attain transformative results.

Feel free to interact with us on the AWS forums or GitHub!