$1.5M+ · AI + Gaming

What they're building

ManaMind deploys autonomous AI agents that play video games from scratch, find bugs, and produce detailed QA reports, all without human testers walking through the game manually. The platform is engine-agnostic: it works purely from captured audio and video, no SDK or game engine integration required, which means it runs across Unity, Unreal, proprietary engines, and mobile builds without any framework changes. The agents navigate menus, enter gameplay, collect evidence of bugs, return to document their findings, and hand off structured reports that plug directly into existing QA workflows. The key differentiator is the underlying model. Sabtain Ahmad, the CTO, built a proprietary vision-language model specifically for game environments after testing models from OpenAI, Google, and Anthropic and finding all of them inadequate at understanding or navigating game UIs. In internal tests, his model outperformed all three on bug detection. THQ Nordic is already using the platform in early access.

Why this matters

QA is typically 12% of a game's total production budget. For a AAA title with a $200M budget, that is $24M spent on humans walking into walls. The scale of the problem is not in doubt. What has been missing is a model that actually understands games, not just images. Existing script-based automation tools require the game to be instrumented in advance and break whenever the UI changes. LLM-based computer use agents fail because they were not trained to interpret the visual language of game interfaces, health bars, inventory systems, dialogue trees, physics interactions, and platform-specific UI conventions. ManaMind's bet is that a purpose-built model for game environments, trained specifically to play and reason about games rather than to complete general computer tasks, is the only approach that actually generalises across genres and platforms. The market timing is sharp: the game industry has laid off tens of thousands of QA testers over the past two years. Studios need a credible replacement, and they need one that does not require months of integration work. Engine-agnostic, audio-video-only operation is ManaMind's answer to the integration problem.