Your company is building a game about a fictitious murder. It’s a “whodunnit” where players need to launch an investigation to crack a murder mystery. The characters are AI-generated. Players interrogate AI-generated suspects, and can even use an AI bot as the interrogator if they don’t want to do the work. The goal of the game is to arrest a suspect.
This game is real. It’s called “Unraveling Shadows” and it was built by AWS Solutions Architect Saubia Khan and Senior Solutions Architect Alexis Philippart de Foy. They finished the first iteration in 2024. While it only took a couple months to build, the process of building a game using GenAI is intricate. Khan and de Foy had to build synthetic data sets to make the suspects harder to decipher and to make games longer. They hosted the front end applications on AWS Amplify and AWS AppSync for backend data retrieval. They also used Amazon Bedrock, Amazon DynamoDB, and Amazon EventBridge.
Khan, de Foy, Hydrolix VP of Product David Sztykman, and Hydrolix Partner Manager Rory McVicar were recently interviewed by AWS about using GenAI to build games and why observability is vital for improving QoE.
“Unraveling Shadows” is an example of the kind of game developers can build quickly and affordably using AWS GenAI services. If they wanted to scale it up to a production game, the amount of data generated would increase astronomically. Think terabytes of log data a day with hundreds of thousands of users playing per second. And, like the gaming industry at large, the content requires real-time streaming. A player takes an action, which leads to another player taking an action, and all of that behavior must be tracked and observed. What type of key stroke is each player making? How fast is the mouse moving? Where are players located? Which devices are they using? For the best quality of experience (QoE), that data needs to be captured, correlated, and analyzed. It’s the only way to spot incidents preemptively, before players are affected.
However, gaming companies struggle with keeping and using all their data, especially when their apps are generating terabytes of logs every day. Many analytics platforms can’t handle the load, and even when they can, the expense to ingest and retain all that data is exorbitant. Many must make trade-offs, putting some data in cold storage which is difficult to rehydrate, or discarding data, which can be essential for identifying the root cause of an investigation. Each data source, whether it’s the CDN, WAF, or any other tool that produces logs, provides a ton of detail on player behavior and incidents. The data is valuable for reducing the mean time to resolution (MTTR) when there are issues, and making business decisions to optimize the QoE for players.
“Unraveling Shadows” is the perfect analogy for why a real-time analytics platform like Hydrolix must be used for observability during the end-to-end gaming experience. The game is based on giving players the information they need to find the murderer. Hydrolix is designed to give engineers, IT operations, and responders every piece of data they need to investigate an issue, whether it’s security or performance related.
Hydrolix can ingest terabytes of log data a day, even a terabyte a minute, compress it 20x-50x for affordable storage, and retain all data for at least 15 months. All data remains hot for rapid querying, and companies can see data insights in less than 10 seconds from ingestion, even when ingesting terabytes of data a day, in one consolidated dashboard. Hydrolix captures AWS CloudFront, WAF, and Elemental data, and can ingest any other immutable, timestamped data, including CDN, SIEM, and application logs. It then indexes the data in real time. Hydrolix extracts insights such as bitrates, video quality, offloading, and ISP issues. Through APIs, SQL, Apache Spark, Elastic and any other connector of choice, LLMs can connect to Hydrolix and ask questions about the data. Even non-technical users can access the data, ask questions, and understand the output.
Imagine having all the facts for an investigation in one place, and being able to query that data in seconds, using the query and visualization tools of your choice. That’s powerful, and makes cracking cases faster and easier.
Next Steps
- Watch the video Unraveling Shadows with Hydrolix.
- If you are interested in trying Hydrolix for free or speaking with one of our experts, visit our Hydrolix for AWS page.
