The Force Multiplier: An Honest Post About AI, Craft, and the Grant Push

Where the grant push stands

Short version: the Epic MegaGrants submission window is close, and the studio is in its final push toward it.

The prototype build has three playable gods with full ability kits. The survival loop runs start to finish, with elite waves, a boss encounter, and a card draft system that now has a proper deck screen. The site you are reading went live this month, along with the Mythology Codex and this Chronicle. Right now the work is capture: recording the build honestly, mechanic legibility ahead of polish, so the application shows what the game actually is.

I am not going to promise a date, because the goals board carries the commitments here, with statuses instead of dates. But this felt like the right moment to write the post I have been meaning to write since the Opening Chronicle. It is about the question I get most, and the one I would be asking too: how does one person with no programming background build all this?

The answer has two halves. One half is AI. The other half is everything AI is not allowed to touch.

What I mean when I say AI writes code here

I am not a code writer by trade. I spent eighteen years in field technology, networking, deployments, architecture, and eventually leadership. That work taught me process, verification, and how to read a system. It did not teach me to write a Gameplay Ability System class in C++.

AI code assistance closed that gap. Here is the honest shape of it:

• AI assistants write a large share of the code in Realms of the Ascended, and I review all of it. My working model is that the assistant is a very fast junior engineer with no memory. I read the diff, I build it, and I play it before it counts as done. The value is in the speed. The ownership stays with me.
• We built our own tooling to make that real. The Forge is our in-house Unreal Engine 5.7 plugin, now more than three hundred tools, that lets an AI assistant work directly against a running editor: read project state, create and edit assets under our conventions, run the game in the editor, and capture its own screenshots to prove a claim. A generic assistant guesses. One with eyes and hands, pointed at your actual project, is a different tool entirely.
• The workflows matter more than the model. Conventions files that load every session. Handoff notes so the project does not restart from zero each morning. A rule that agents test their own work and bring evidence before a human looks. The report is not the result. The result is the result.

None of that is magic. It is the same discipline I ran field teams with for years, pointed at a new kind of teammate.

Taking the "AI slop" concern seriously

I know the term, and I am not going to pretend the concern behind it is not real. Stores are filling with low-effort, AI-generated games. Artists are watching tools trained on their work undercut their livelihoods. Players are right to be tired of products where nobody cared enough to check the output.

Those concerns deserve a straight answer, not a defensive one. Mine is this: slop is a decision, not a property of the tool. Slop is what happens when generation replaces judgment, when nobody reviews, nobody plays the build, nobody says no. The tool does not remove the responsibility to care. It just makes carelessness faster.

So the question I hold myself to is not whether AI touched the work. It is whether the work would survive scrutiny from someone who cares about the craft:

• Every system in this game exists because I designed it, argued with it, and played it. The design document predates any of the AI tooling by years.
• Every line of generated code gets read, built, and run before it counts. Claims get measured, not trusted. We have thrown away entire deliveries that did not meet the bar, and I have written here before about what that cost.
• Nothing ships to this site or the build because a model produced it. Things ship because they passed a gate a human set.

There is a real difference between AI as the product and AI as a tool a person uses to build their product. The first is slop by default. The second is just leverage, and leverage has always been the story of independent development: engines we did not write, marketplaces of assets we did not model, middleware for everything. AI is a new rung on a very old ladder.

Where I do not use AI

This is the part I want to be loudest about. For 3D modeling, rigging, animation work, and the character art that defines how this game looks, I do not use AI. Not because I could not, but because I have quality standards and human skill is still clearly ahead where precision matters. A rig that deforms correctly, a silhouette that reads at isometric distance, topology that animates cleanly: that is craftsmanship, and I pay for craftsmanship. The one exception is concept art exploration, where I use AI the way you would use a mood board, to communicate direction before any real work begins.

The pipeline is simple and it works: freelance platforms let a self-funded studio hire world-class specialists one milestone at a time. Thor and Odin, along with Odin's spear Gungnir, were modeled by Serhii Prytula, whose work you can also find on Fiverr. The models are excellent, and as of this posting he is sculpting a custom Mjolnir for Thor. When it is finished, the renders will go up on Thor's compendium entry so you can judge the craftsmanship yourself. Scoped milestones, visible checkpoints, work delivered into a repository I control. Every lesson from the contract that went wrong last year, applied to contracts that go right.

That is the budget philosophy in one line: spend the money where humans are irreplaceable, and use AI to afford it.

The force multiplier argument

Here is the perspective I want to offer anyone weighing in on AI and games, and especially anyone like me who has a design they believe in and no studio behind them.

Two years ago, a person in my position had three options: learn programming for some years first, raise money against a document, or contract the build and hope. I tried the third and wrote publicly about how it went. What AI changed is that it added a fourth option: build the prototype yourself, with an assistant that types faster than you and a discipline that keeps the quality yours.

Here is what that fourth option looked like in practice. The clean restart was in February. From February to June, every hour of progress on this game came from late nights and weekends, worked around a high-stress director job. The workflow is the reason those hours compounded instead of evaporating: with handoff notes and a scoped queue, a Tuesday midnight session picked up exactly where Sunday left off. In June I left that job to commit to this full time. Six months from restart to a playable three-god prototype, procedural dungeons, a survival mode, an in-house tooling plugin, and a live studio site. I updated the goals board alongside this post so the shipped history now reaches back to that February restart. It is not a list I could have produced alone with any amount of caffeine. It is what one focused person plus a force multiplier plus a system for never losing progress can actually do.

That is what a force multiplier means. It did not design my game. It did not have the idea, and it does not care whether the idea ships. It multiplied the effort of the one person who does. The result is a playable prototype, built for a fraction of what a contracted build costs, owned line by line by the studio that made it.

And the goal of that prototype is not to stay a solo AI-assisted project forever. It is the opposite. The point of the grant application, and of the funding path beyond it, is to earn the resources to hire professional programmers and Unreal specialists, to grow from prototype to demo to 1.0 with a real team. AI got this studio to the starting line. People will carry it from there. More working artists and engineers employed, not fewer: that is what this tool made possible for us, and I do not think we are the only ones.

If you are a designer, a writer, a systems thinker with a game you have carried for years: the barrier between you and a playable prototype has never been lower. The standards you hold yourself to still decide everything. The tool will amplify whatever you bring to it, care or carelessness alike. Bring care.

What comes next

The goals board has been updated alongside this post with where everything stands, including the capture work and the application itself. When there is news on the grant, you will read it here first.

Thank you for following along. If you want these posts in your inbox, join the Ascended. If you disagree with anything I wrote about AI here, I genuinely want to hear it: info@viseterna.com reaches me.