How I Cut Video Game Research Time by 80% with Amazon Nova Act
Building a personal AI Agent that turns hours of browsing into minutes of action
Banjo Obayomi
Amazon Employee
Published Apr 1, 2025
Last month, I spent an entire evening researching the best PlayStation 5 games to add to my collection. Two hours of opening dozens of browser tabs, jumping between review sites, checking prices, and comparing features, only to realize I'd forgotten to note which games supported multiplayer. I had to start over for several titles. This kind of tedious web research is the perfect example of a task that's necessary but inefficient.
Amazon Nova Act has completely transformed this process for me. I've built a tool that now completes the same comprehensive game research in just 15 minutes, automatically gathering and organizing all the information I need to make informed purchasing decisions.
Amazon Nova Act is an SDK that lets developers build agents capable of performing actions in web browsers—essentially digital assistants that navigate websites the same way a human would. Available as an early research preview, this tool converts complex web tasks into a series of simple, reliable commands.
What makes Nova Act particularly useful is how well it integrates with Python. This means you can combine browser automation with data processing capabilities, perfect for research tasks where gathering and analyzing information go hand in hand.
If you're a gamer, this process probably sounds familiar:
1. Visit GameFAQs or Metacritic to identify highly-rated games for your system
2. Open Amazon or other retailers in separate tabs to check current prices
3. Search for detailed descriptions and specifications for each game
4. Check user reviews to get real player perspectives
5. Make notes comparing features, prices, and ratings
6. Repeat for every game you're considering
For someone researching 10-15 games, this process easily consumes 1-2 hours. And if you want updated information a week later (perhaps prices have changed), you need to start from scratch. It's a perfect candidate for automation.
Using Amazon Nova Act, I've built a Streamlit application that automates the entire game research process. Here's what it does:
- Finds top games for any selected gaming system on GameFAQs
- Searches Amazon in parallel for each game to find prices and descriptions
- Compiles results into an interactive table with all the information
- Saves all research for future reference or comparison
The best part? This entire process now takes about 3 minutes, with minimal user input required. Let's look at how it works behind the scenes.
The first phase involves navigating to GameFAQs and extracting information about top-rated games:
What's fascinating about this code is that we're telling Nova Act what we want in natural language ("Click on the link for..."), while also receiving structured data through schemas. This balance of flexibility and precision makes the tool exceptionally powerful.
Once we have our list of top games, we search Amazon for each one simultaneously using parallel processing:
For each game, Nova Act opens a new browser instance:
This parallel approach is what enables the dramatic time savings. Instead of searching for each game sequentially, we're researching all of them at once.
After using this tool for my game research, I've seen several concrete benefits:
- Dramatic Time Savings: Research that took 1-2 hours now completes in around 15 minutes (an 80-90% reduction)
- More Thorough Comparisons: I now compare more games before making decisions (15 versus the 5-6 I could manage manually)
- Better Purchase Decisions: With comprehensive data at my fingertips, I've made more informed choices
- Price Tracking: Running the tool weekly lets me track price changes and grab games when they go on sale
- Research Organization: All my research is saved in a structured format that I can review anytime
The biggest surprise was discovering several highly-rated games I'd never heard of, simply because I could now efficiently research beyond the obvious popular titles.
While I've focused on game research, the same approach can be adapted for many other areas where web research is typically time-consuming. Here are some ideas you might want to explore:
- Travel Planning: Create an automation to compare flight prices, hotel reviews, and local attractions simultaneously when planning trips. This could help you discover boutique accommodations that don't appear prominently on major travel sites.
- Product Research: When shopping for electronics or appliances, you could build a version that researches specifications, prices, and reviews across multiple retailers at once, potentially uncovering better deals or configuration options.
- Real Estate Monitoring: Set up a daily automation to check housing listings in neighborhoods you're interested in, capturing new properties and price drops before they get wide attention.
- Academic Research: For students and researchers, a tool that searches academic databases and organizes papers based on relevance, citation count, and publication date could dramatically speed up literature reviews.
- Job Market Analysis: If you're job hunting or considering a career change, automating searches across job boards to track salary ranges and required qualifications might help identify emerging trends in your industry.
- Financial Research: When considering investments, an automated process could gather analyst ratings, recent news, key financial metrics, and competitive comparisons all in one consolidated report.
These are just starting points - the framework is flexible enough to adapt to almost any research-intensive task. The key is identifying repetitive web browsing patterns that you can translate into Nova Act commands.
If you're interested in building a similar tool for your own research needs:
1. Visit https://nova.amazon.com/act to generate an API key
2. Install the SDK with `pip install nova-act`
3. Set up your authentication by exporting your API key as an environment variable
The code examples above provide a starting point, but you can adapt the approach for virtually any web research task you currently do manually.
When I first built this Nova Act solution, I was simply trying to save myself some time while researching games. What I didn't expect was how it would fundamentally change my approach to online research in general. Tasks I previously avoided due to their tedious nature are now handled effortlessly through automation.
The power of Amazon Nova Act isn't just in replicating human web browsing—it's in doing it at scale, with precision, and in a way that integrates with data processing workflows. For anyone who regularly conducts web research of any kind, it represents a significant leap forward in productivity.
As browser automation tools like Nova Act become more accessible, I expect we'll see a shift in how people approach information gathering online—moving away from manual browsing and toward automated research assistants that handle the tedious parts while humans focus on decision-making.
Now go build!
Any opinions in this post are those of the individual author and may not reflect the opinions of AWS.