Executive Summary

Roboto Games achieved 50% cost reduction by implementing nunu.ai's automated testing solution. The integration required minimal dev time (<5 hours) while delivering comprehensive testing coverage across their multiplayer survival game Stormforge, resulting in 160 reclaimed QA hours monthly and 400+ automated tests.

About Roboto Games

Roboto Games is a 13+ person game studio developing Stormforge, an open-world survival crafter built with Unity. The studio is led by Curt Bererton and Mathilde Pignol, who previously managed the PopCap San Francisco office together, where the team worked on hit franchises Plants vs. Zombies., Peggle, and Bejeweled.

Onboarding and Integration

Our partnership began with a two-week onboarding where Roboto Games integrated our SDK and learned to use our 'nexus' web portal. With our hands-on support, they spent less than five hours of dev time on integration. Within two weeks, we successfully launched our first automated AI agent tests. Together, we expanded test coverage and enhanced agent capabilities over time.

How is nunu.ai helping Roboto Games?

Once integrated, our tool allows Roboto Games to create test cases in pure natural language (e.g. “Craft a pickaxe” or “play the tutorial”). Our AI agents automatically execute these tests daily or with each new build. For every executed test, we provide a gameplay recording with all necessary details to ensure reproducibility: The full reasoning of the AI agent, the exact actions it took (Keyboard/Mouse/Controller input), error logs and a test report. The team receives instant notifications about bugs through their preferred platform (Slack, Jira, etc.). Through nunu’s web portal (nexus), everyone on the team - from developers to artists - can create and manage test plans without technical expertise.

Functional and Regression Testing

Roboto Games conducts daily tests covering a wide range of Stormforge's most important systems, including crafting, building, combat, inventory management, and character creation. Each night, nunu.ai automatically pulls the latest game build and runs these tests (totaling over 50 test steps) to ensure everything still works as expected. If a test fails, Roboto Games is automatically notified and can review the recording or report for further details.

Using nunu.ai, Roboto Games has identified bugs such as:

• “Naming the character is broken.”
• “Splitting items doesn’t work.”
• “Buildings get damaged over time when they shouldn’t.”
• “Item description bug”

… and more

Here is a small excerpt of such system tests:

Performance Testing

In addition to daily test runs, Roboto Games runs performance tests for every single game build (averaging 15 builds per day). During these tests, the AI agent starts a new game, creates a new world, and explores it. Throughout the run, nunu.ai tracks key metrics such as:

• FPS
• CPU/GPU usage
• Memory utilization
• Frame time

This approach allows Roboto Games to immediately identify changes that lead to performance issues, memory leaks, or excessive GPU/CPU usage. Furthermore, nunu.ai conducts these tests on consumer hardware providing insights into how the game performs on end-user devices.

Results

• 50% cost reduction compared to outsourced QA contracting rates.
• Fully automated QA processes, eliminating manual testing overhead. Robot Games exclusively uses nunu.ai to manage their entire QA needs.
• Reclaimed 160 hours per month by automating tedious and repetitive tasks.
• Increased testing coverage, with over 400 monthly tests (including performance tests), ensuring core systems work reliably. This has improved their iteration speed and time to market.

Looking forward

While our AI agents can fulfill a broad range of tasks, such as crafting items, navigating UIs, performing basic combat, and placing workbenches, they still have their limitations: It does not yet excel at tasks like constructing large coherent buildings or discovering optimal boss fight strategies. We are constantly working to improve agent capabilities and increase their test coverage overall. Book a call to learn more about current capabilities and what the future holds.