Title
Introducing Aether Nexus and Aether Forge: A UE5 AI Prototype Pipeline
Forum Post
Hey everyone,
I wanted to introduce a project I’ve been building called Aether.
Aether is my attempt to make AI-assisted Unreal Engine development less chaotic and more useful for actual prototype work. I am not trying to sell this as a magic button that turns one prompt into a finished commercial game. That kind of pitch sounds good for five minutes, then falls apart the second you need structure, repeatability, source control, debugging, or a real UE project that still makes sense afterward.
The idea behind Aether is simpler and more practical:
Take a game idea, turn it into a structured plan, hand that plan to Unreal in a controlled way, and keep a clear record of what was created.
Aether is split into two parts.
Aether Nexus
Aether Nexus is the planning and orchestration layer.
Nexus is meant to understand a game development prompt, break it into a recipe, decide what systems and assets are needed, manage provider or API connections, stage files, prepare manifests, and decide what should be handed to Unreal.
It is the command center.
It does not directly mutate the Unreal project.
Aether Forge
Aether Forge is the Unreal Engine 5 execution layer.
Forge is the UE5 plugin side. Its job is to receive approved work from Nexus and perform editor-side actions such as creating folders, levels, Blueprints, imports, run structures, and validation-oriented reports.
It is the build side.
The short version is:
Nexus plans.
Forge builds.
Unreal validates.
Why I’m building it this way
A lot of AI tools can generate text, code, images, or isolated assets. Some can help with Blueprints or editor actions. The problem I kept running into is that the workflow still feels scattered.
You can end up with one tool for planning, one for assets, one for textures, one for dialogue, one for Unreal tasks, and then a pile of manual cleanup afterward.
That is useful in pieces, but it does not feel like a production pipeline.
Aether is being built around a cleaner flow:
Prompt
→ Nexus recipe and plan
→ provider or local asset staging
→ Forge action manifest
→ UE5-side execution
→ validation and evidence report
→ revision or repair pass
The important part is the handoff.
I do not want a general AI agent freely editing an Unreal project with no structure. AI-driven editor automation can get messy fast. So the architecture is being built around manifests, allowlisted actions, run-scoped output folders, provider staging, and evidence reports.
That boundary matters:
Nexus prepares the work. Forge is the only side that should touch the Unreal project.
Current development focus
Right now this is still in private MVP development. I am focused on proving one clean, repeatable end-to-end workflow before I try to put anything on Fab or call it a product.
The first target vertical slice is:
Prompt → dark fantasy third-person boss arena prototype foundation
For that first slice, the goal is to assemble or generate things like:
project and run folder structure
level shell
player/controller setup
basic Blueprint scaffolding
enemy or boss Blueprint structure
health and damage loop
materials and assets from staged manifests
validation report
revision flow
The goal is not to make the most beautiful boss arena ever. The goal is to prove the pipeline:
Can Nexus plan it?
Can Nexus stage the needed work?
Can Forge receive the manifest?
Can Forge build the UE-side structure?
Can the system report what happened clearly?
Once that loop is reliable, expanding the recipes becomes much more realistic.
Local-first direction
Aether is being designed to support local-first workflows where possible.
Some of the integrations I am planning or staging around are:
Ollama for local LLM planning
Blender Python for procedural placeholder meshes
ComfyUI for image, icon, or texture-style generation
local Kenney or Quaternius asset libraries
optional cloud providers later
optional runtime AI or NPC systems later
The goal is not to force users into paid API chains just to test the tool. Cloud providers should be optional. Local tools and local libraries should be useful on their own.
What I’m looking for feedback on
I’m posting this before any marketplace release because I would rather get honest feedback from Unreal developers early than polish the wrong thing.
A few questions I’m trying to answer:
Would a Nexus plans / Forge executes workflow be useful to you?
What UE5 actions would you want safely automated first?
Would you rather see this as a prototype-generation tool, a general editor assistant, or a pipeline framework?
What would make you trust an AI-assisted UE5 tool enough to use it in a real project?
What should absolutely not be automated?
Does local-first support matter to you?
What kind of demo would convince you this is worth using?
Important clarification
Aether is not being presented as a finished commercial game generator.
The intended use case is:
AI-assisted UE5 prototype acceleration with structured planning,
controlled execution, and transparent reporting.
I am trying to keep this grounded. The goal is not to replace developers. The goal is to reduce repetitive setup work, make prototype creation less scattered, and give AI-assisted Unreal work a cleaner pipeline around it.
If there is interest, I can share updates as the private MVP matures, especially once the first full Nexus to Forge demo is ready.
Thanks for reading. I’d appreciate any technical feedback, skepticism, or feature suggestions.
DISCLAIMER
I have a hard time putting words to paper so i typed up a very long word doc and had AI translate my thoughts for this post. So please, don’t embarass me with pointing out the obvious. I have a hard time being on a forum as is, but I found something I’m super passionate about and wanted to share it with the world.