Four-day Unreal beaver game: creator shows ten AI tools, Rodin rigging and Mixamo fixes
By GameSlash
Lao Liu reveals an Epic AI MegaJam workflow for a beaver life sim: different agents code the game while Rodin, Mixamo, Suno and ElevenLabs handle assets.
On November 19, 2026, Lao Liu published a six-minute account of his four-day Epic AI MegaJam 2026 build: an Unreal Engine beaver life sim with dam upgrades and an underwater settlement. He names ten AI tool instances, shows gameplay interactions and openly describes what failed. The production time and team are reported by the creator. Original development video.

A real player loop before visual polish
The beaver digs mud, builds and upgrades a dam, and expands homes underwater. New animal neighbours reward house construction with better tools; land exploration introduces wolves, bears and eagles. Lao Liu lists two GPTs, three Claude instances, Cursor, GrokBot, Rodin, Suno and ElevenLabs. The language agents handled gameplay rules, UI and saves, while specialist services produced assets.

Why rigging made the creator switch tools
Rather than spend four days making the world from scratch, he reused a stylized environment pack. Rodin produced characters, materials and skeletons through MCP, but making those characters move convincingly was harder. After struggling to generate animation, he selected existing Mixamo motions and had an agent retarget them in Unreal. Suno handled music and ElevenLabs sound effects.

What did not ship
He acknowledges unfixed bugs, a surface map laid out in code without proper level design, and a house entrance without a finished interior. He says the prototype still lacks fun. The lesson for developers is to focus model time on a solid interaction loop, character motion and playtesting, and use suitable pre-made assets when the schedule is tight. GameSlash has not independently played this build.
Read the original creator post and the Epic AI MegaJam page.