AWS Logo
Menu
STAR-EX AI intro with Amazon-Q

STAR-EX AI intro with Amazon-Q

Part 1 building new idle gameplay in STAR-EX world, collecting resources from across blockchain, then craft them to advance or sell them into market.

Published Jan 16, 2025
Part 1: STAR-EX intro and Amazon-Q feedback
Part 2: Fullstack development with AWS tools
Part 3: Amazon bedrock
Hi everyone, I'm Dale Watson, an 8y fullstack dev. All I can say is I'm so happy knowing Q and want to share my story. This is the best DX ever. I've been experiencing improvement and seamlessness in my workflow.
Something that makes me addicted to using it, and it's way better than ChatGPT. The reasoning for development language seems higher here in Amazon-Q, although I'm still unsure what LLM Amazon-Q uses. I don't even need to copy-paste everything; Amazon-Q scans my repo easily, and I can learn more here with Q than with any other AI agent I have tried so far.
Yes, there are some limitations in Q, but if you know the prompt and are able to teach Q, it will make the code high quality. My suggestion: I think simply step by step, which I will show you in Part 2.
Amazon-Q also helps me easily by teaching me how to use other Amazon services. I was really overwhelmed with how many tools there are, as I needed to open docs one by one, getting redirected to multiple links. It still annoys me how to even start—like, which article should I see, which blog, video, or example should I check? And the AWS website is really old-school and uninteresting, alongside those boring video explanations. But damn, Amazon-Q makes me want to learn them all because it goes well with my preferences. Thankfully, I already understand the latest Amazon services now and know exactly what I can use next.
And I'm someone who wants to see right away how to do it rather than only logically. So with Amazon Q, they give me the steps right away to do it in my OWN PREFERENCES. Yes, my own preferences, which shape the guide-steps into what I'm really looking for. So if I'm just looking at how to set it up and have a question, I can get the answer immediately. Compared to reading articles, where there are questionable statements and I can't get an immediate response.
So because of that, I'm currently back in my learning & experiment journey with Amazon to learn all the products that I will use with this game STAR-EX, which I'm already using some of.
Current Implementation (6 AWS Tools in Use):
  • Backend: AWS DynamoDB, AWS Lambda, AWS RDS
  • Frontend: AWS Amplify, AWS S3
  • AI tools: Amazon-Q for enhanced development experience
Ongoing Development & Learning (5 AWS Tools):
  • AI Agent Development Goals:
    • AWS Bedrock: To build an AI agent, including an in-game navigator and assistant.
    • TITAN & NOVA LLM: Experimenting with these cutting-edge models.
  • Real-Time Multiplayer: Using AWS RDS for backend setup.
  • Image Retrieval: Leveraging AWS Rekognition.
Future Plans (4 AWS AI Tools):
  • AI Governance Systems:
    • Employing AWS Nova and AWS Kendra for Retrieval-Augmented Generation (RAG).
    • Exploring AWS SageMaker Canvas for no-code AI prototyping, with plans to dive into SageMaker for advanced model training.
  • Predictive Balancing: Using AWS Forecast to analyze player data and adjust the in-game economy.
  • Custom AI Model Training: Developing AI for governance and world updates, ensuring continuous growth.
So with 15 AWS tools and services, with plans to expand further as I train models for AI governance. I can’t wait to see where this journey takes STAR-EX.

In the part of the AWS game hackathon journey, (link to devpost) I make a new IDLE gameplay which we call it EXPEDITION, even though I made it late in the last 2 weeks due to my other work project. This is still sneak-peak demo that soon will improve a lot. In the last months, I made a game concept and developed its first arcade gameplay, which is Exploration.
STAR-EX 3D graphic arcade
Exploration gameplay - dessert
STAR-EX 3D graphic arcade
Exploration gameplay
Garage
This franchise game called STAR-EX. Imagine a cross-chain, cross-platform, cross-dimension, cross-story, cross-gameplay, and cross-community experience as the ultimate entertainment for blockchain. We're creating a GameFi ecosystem that leverages multiple blockchains, where their ecosystems either compete against each other or collaborate, aiming to become the next-generation Formula 1 (F1) of Web3 esports. It’s more than just competition; it’s blockchain entertainment. Players won’t need to worry about choosing a network chain—they can focus on our dynamic "war-system" that drives adrenaline and fuels the hype. New players can either learn indirectly from the esports events or immerse themselves in the rich lore and story behind the game.
Main Factions for each Popular blockchain
Well, but more than that, after the span of these weeks and learning AI, I wanted to leverage the game using AI and ML. So, the main resources system will be balanced and adjusted by AI because this will be a complex game—imitating the real world and being realistic while combining it with fantasy. It will include a governance system for each zone, a currency system, finance, and economic power to sustain the world ecosystem and the game. That's why it will be connected to blockchain to make the ready system market as well, instead of setting them all up again here.
The general idea is the game will provide economic warfare and improve until we can call it an AI-UNIVERSE. So, the AI will take care of the next updates and constantly grow to develop self-sustainability, increasing both perfection and complexity. I have an architecture plan so the game can scale modularly.
Besides that AI-UNIVERSE, I'm planning to use multiple AI-agents. Either to help players easily understand the game by connecting the AI-agent to the knowledge base of my documentation, so players can just find out what is the combination of these items, what is the term everyone is talking about, what about this UI, or what about the current price in the market. Everything that can make players easily play or even co-pilot the gameplay effortlessly.
We have this in our next gameplay, which is IDLE gameplay. Yeah, in this hackathon, I started the next gameplay, which I call Expedition. Players should be able to easily co-pilot and command the AI to do their tasks seamlessly—that's my goal. I will share more about this in my next blog, PART 2 & 3, to give readers a sneak peek of those features.
Again, STAR-EX is designed to feature three core gameplay modes, each utilizing the same shared resources to create a seamless, interconnected experience:
  1. EXCEED RACE – A competitive, high-speed race across various sectors of the galaxy.
  2. EXPLORATION – An intense arcade, resource-gathering mode that allows players to discover new planets and gather valuable materials.
  3. EXPEDITION – A strategic idle mode where players embark on tactical missions, manage resources, and expand their influence across the STAR-EX universe.demo EXPLORATION: STAREX.APP
    demo EXPEDITION (current hackathon): real.d2mqz44b8dldbz.amplifyapp.com
    social media: youtube [discord
    Sneak peak EXPEDITION
    ](https://discord.gg/starex)

So what's Amazon-Q that I used? It's a VS Code extension that can scan my repo, understand the situation, and help accordingly to my requests. It can generate a lot of code that most other AI agents would consider "too much," but there's no such thing as "too much" in Amazon-Q. Instead, it asks you follow-up questions and gives pinpoint suggestions about what’s next.
You can also use it to generate testing from your codebase, update your docs, or of course, refactor. This improves development and ensures high-quality code. The reason most projects have crap code is because it’s too tiring to write everything, analyze it, and maintain it. But what if now there’s AI that can help us?
I don’t need to give instructions on how to install or provide more details about Amazon-Q—you can find that elsewhere. Instead, I’ll share my journey, feedback, and insights so you know what to expect.
Let me share the UNFULFILLING experiences first, which are actually negligible but would be huge upgrades—as well. Hopefully, the team can see this feedback.
1. The chat widget is taking over the folder widget [IMPORTANT]
This means I can't see both the chat and my folder simultaneously. Maybe the chat could be on top? Right now, I always need to press cmd + option + i (to open the chat page) and then go back to the folder page (cmd + shift + E) to select my files.
2. Chat widget resets the scroll position [IMPORTANT]
Every time I go back to the chat widget, it always scrolls back to the top, even though I’ve had many conversations.
3. Limited integration for generated code
While we can copy the code or directly click the "cursor" to add it, I’d prefer something more powerful—like letting the AI generate the files or even replace code lines directly. Right now, it’s still manual.
4. No search feature in chat [IMPORTANT]
There’s no way to search for specific text in the chat. This makes it hard to trace back what I’m looking for, especially if I’ve had many conversations.
5. Cannot close chat tab with the keyboard
The chat tab cannot be closed using a keyboard shortcut.
6. Slow document generation
Generating documentation takes too long. When done with the correct prompt, it analyzes the whole codebase, which takes a long time.
7. Lack of memory within conversations
It doesn’t remember tasks it’s already done, the packages or libraries we actually use, or even the ones it recommended earlier. For example, I once mentioned that I don’t use nextjs and told it to stop recommending router-link for that framework. But after a while in the same conversation, it repeated the recommendation.
8. No voice command or voice-to-text capabilities
There’s currently no support for voice commands or voice-to-text functionality.

STRENGTHS
Well, that’s still not stopping me from enjoying Amazon-Q’s strengths and features. I really like:

1. The way Amazon-Q questions back

I LOVE how Q keeps questioning me at the end to make sure everything’s good or to extend its knowledge base, becoming smarter each time. This is exactly how people learn—making predictions and ensuring an all-knowing approach. That’s why I know it’s the BEST for Developer Experience (DX). ChatGPT doesn’t even do this, which shows it lacks new graph-learning networks.

2. Session management and tab handling

I appreciate that Q can open a new tab for different cases. It doesn't mix up the sessions between tabs, and they even auto-close inactive sessions after 24 hours. However, I’m not entirely sure if it remembers the conversations from those closed sessions.

3. Quick docs generation

Despite the speed issue I mentioned earlier, it’s still incredibly useful for generating quick documentation and even helping with code updates for commits and more.

4. Inclusion of source references

The fact that it provides source references is AMAZING. (I think this is part of its latest updates, and it’s a game-changer.)

5. Multi-language support

While Amazon-Q states it’s mostly focused on Java, I’ve seen its capability with React and other languages. Sometimes, its understanding of React feels even better. That’s why I’m eager to use Q to learn other Amazon tools or gain general knowledge faster.
As a developer, I also love exploring languages like Rust, Golang, Move, and Cairo for smart contract development. It feels like Q is capable of helping me dive into those areas too.

Again, there’s almost no chance I’ll stop using Amazon-Q. I’m fully invested in creating high-quality projects and even cleaning up older ones. PART 3 might get delayed, but it’ll be worth it.
Stay tuned for what I’m building fullstack in PART 2!
 

1 Comment