A practical workflow for AI game development

What is an AI-native game engine?

An AI-native game engine makes AI assistance part of the game creation workflow. In Makepot, that means working with a 3D scene, project assets, and an AI assistant in the same workspace. You describe an intention, examine the resulting changes, and decide how the project should evolve.

Makepot is a product of Arclusion, a computer science research lab focused on simplifying 3D development. It is currently in early access preparation. The workflow below describes how to approach a small game project and what to check as you iterate.

1. Start with one playable idea

Choose a player action and a clear goal before planning a large world. For example: explore a small island and reach a lookout point. Decide how the player moves, what they need to find, and how they know they have finished. A short playable loop gives you something concrete to test.

Write down the scene, the action, and the mood separately. A useful brief might be: “A small island exploration game with a path from the beach to a hilltop. Keep the route easy to follow and leave space around the starting area.” Treat this as a starting brief that will need iteration.

2. Build and inspect the 3D scene

Work on the layout before adding visual detail. In the Makepot editor, use the scene view and project assets to inspect what is in the world. Look at the relationship between terrain, objects, and the route a player will take.

When working with an AI assistant, ask for a change you can evaluate. Adjusting a starting area or refining a path is easier to judge than asking for every part of a finished game at once. Keep the intended player experience in the request.

3. Refine changes in your project

AI assistance is most useful when you can review its work. Makepot changes land in your project, where you can inspect, tune, or undo them. Check whether a result supports your idea before building the next layer on top of it.

Separate visual checks from gameplay checks. A scene can look convincing while a route is too steep, a goal is unclear, or an important object is hard to see. Refine the part that affects play, then revisit the appearance.

4. Playtest the experience

Play through the route from the starting point to the goal. Check movement, camera framing, collisions, and whether a new player can understand what to do. A successful build is only one checkpoint; the game still needs to feel understandable and responsive.

Ask someone unfamiliar with the project to try it. Observe where they hesitate and which actions they attempt. Use that feedback to make a small, specific revision, then play through again.

5. Prepare a browser release

Makepot includes browser-game publishing so you can share a link with players. Test the published version as well as the editor preview. Load it on the browsers and devices you intend to support, and check the first load, controls, and the complete playable loop.

Large textures, complex scenes, and too many visible objects can affect loading and responsiveness. Keep the first release focused and expand after people can reliably play it.

Get access to Makepot

Join the early access waitlist for product updates. Public subscription pricing and AI credit allowances are still being finalized; read the current pricing details. For product questions, visit Makepot support.