AI-native architecture
Built around explicit, machine-readable structures that AI coding agents can understand, create and modify.
Meet Konjur. An AI-native game engine built in Rust, designed to put creativity first and let developers — and their AI tools — work in harmony.

Traditional engines were designed for humans clicking through complex editors. Konjur starts with a different question: what if an engine were built from day one for people and AI to create together?
Built around explicit, machine-readable structures that AI coding agents can understand, create and modify.
Keep game content declarative and inspectable, separating what a world is from how the engine runs it.
A modern systems-language foundation with a focus on performance, safety and predictable behaviour.
Designed for reproducible, script-friendly workflows that fit naturally into an agent-assisted development loop.
A visual window into your world, without turning the editor into the centre of the entire creative process.
A focused engine philosophy: composable foundations, clear interfaces and room to grow.
The engine should help you understand your game — not get between you and the game you're imagining. Konjur aims for a practical blend of structured data, straightforward tools and useful visual feedback.
A development process where human intent, structured game data and AI-assisted code changes can work together instead of fighting each other.
Express your scene, mechanics and systems in a form both humans and coding agents can reason about.
Use AI-assisted workflows to produce and evolve content and logic, with explicit outputs you can review.
Run, inspect, test and iterate. Stay focused on the experience you want players to have.
# A world should be understandable to humans and agents world: "the_first_level" entities: - name: "player" components: [transform, controller, camera] - name: "platform" components: [transform, mesh, collider]
Konjur is taking shape as a lean, AI-native game engine. Follow the journey as the architecture, tools and workflows evolve.
Back to the beginning