Uploaded April 2025 | Updated September 2026, 2 weeks ago
As games grow increasingly complex, they present significant challenges in QA and data parsing throughout development and post-release. Watch the full replay of this GDC 2025 session to explore two critical solutions within the development pipeline. First, we explore how Visual Language Models (VLMs) can efficiently detect bugs in expansive open-world games, where manual inspection becomes impractical. Next, we discuss how retrieval augmented generation (RAG) with large language models (LLMs) can summarize the vast amounts of data generated by these systems, enabling developers to quickly identify key insights and address the most urgent issues in their games.
See NVIDIA's announcements at GDC 2025: nvidia.com/en-us/events/gdc
Get started with NVIDIA tools and resources for game developers: developer.nvidia.com/industries/game-development
As games grow increasingly complex, they present significant challenges in QA and data parsing throughout development and post-release. Watch the full replay of this GDC 2025 session to explore two critical solutions within the development pipeline. First, we explore how Visual Language Models (VLMs) can efficiently detect bugs in expansive open-world games, where manual inspection becomes impractical. Next, we discuss how retrieval augmented generation (RAG) with large language models (LLMs) can summarize the vast amounts of data generated by these systems, enabling developers to quickly identify key insights and address the most urgent issues in their games.
See NVIDIA's announcements at GDC 2025: nvidia.com/en-us/events/gdc
Get started with NVIDIA tools and resources for game developers: developer.nvidia.com/industries/game-development
