Announcing Quickscope - tools for automating smoke tests

Welcome back to the Regression Games newsletter! Once a month, we’ll bring you the latest AI + gaming news, resources, and updates from the Regression Games team.

Join our community on Discord and checkout out our tools!

Product Launch - Quickscope, a tool for automating smoke tests in games

Regression Games is excited to release the next iteration of our game QA automation platform, Quickscope. Our new product allows developers to quickly automate smoke tests within Unity, with no code required.

🤖 🦾 Recording and playback tools - Use the in-game overlay to record a gameplay session in-editor or within a build, which can later be played back as an automated test. The tool doesn’t just play back inputs at the recorded timing - the tool also takes loading latency and other dynamic game features into account.

🎥 🎮️ Deep extraction of game state - Our unique AI agent, testing, and recording tools don’t utilize custom code or ray-casting to grab state from the game. Instead, we automatically extract GameObject information from the scene, including fields and properties on your objects. No more pesky agent state code!

🧪 🐛 No-code Test Builder - Once a recording is made, the extracted state, screenshots, and input information can be used to build flexible tests in our no-code web editor. Easily design a suite of functional tests to ensure you’re releasing the best game possible.

Trello Board of Resources in Automated QA

Looking for a collection of great resources to learn more about automation in game QA? This Trello Board has a ton of articles, tutorials, talks, and tools for getting started, whether it be bots, general advice, CI pipelines, and other topics in this space.

Link: https://trello.com/b/nGE5yqZk/game-automated-testing-resource-hub

Papers we’ve been reading

With base LLMs getting more powerful, we’ve been seeing more papers come out with practical engineering tricks to get more usable outputs. The paper Many-Shot In-Context Learning from Google DeepMind discusses the benefits of adding many examples (on the order of hundreds) to increase model accuracy without fine-tuning.

Members of our community discuss these papers on our Discord #reading-club channel. You can also find all past resources on our GitHub.

Have feedback about our newsletter? Let us know at [email protected]

You are receiving this email because you registered on our mailing list or are a developer on our platform.