Why Exodus-Protocol exists

An experiment in how games change when AI becomes part of how they are played, how they are built, and how what happens inside them becomes entertainment.

Exodus-Protocol is more than an attempt to build another space strategy game.

It is an experiment in how games can change when artificial intelligence becomes part of the way we play them, part of the team that creates them, and part of the infrastructure that turns what happens inside them into entertainment.

We are deliberately combining ideas that have already proven themselves with technologies and development methods that have only recently become practical.

The objective is not simply to create one game with AI features. We want to understand whether AI changes how games are played, how ambitious games can be built by smaller teams, and how the technology created for one game can become the foundation for many others.

A proven rhythm, a new interface

Persistent strategy is not a new idea.

Games such as OGame demonstrated more than two decades ago that players are willing to build civilizations over weeks, months, and even years. OGame has been running since 2002, still counts its audience in the millions, and describes a typical interaction as only a few minutes while fleets and long-term progression continue for hours or days.

What has changed is how people interact with digital products and games.

Research into idle games describes a shift from playing to planning: the player makes decisions, configures what should happen, and allows the game to continue with limited attention. A study of 66 idle games specifically identified player attention itself as a resource.

Modern mobile-game behavior points in the same direction. GameAnalytics' 2026 benchmarks show median mobile sessions of roughly 3.1 to 3.5 minutes, while players return approximately 3.8 times per day.

We believe persistent strategy is unusually well suited to that behavior.

You check what happened. You make a few meaningful decisions. You set things in motion. Then construction, research, mining, and fleet movement continue while you work, study, or do something else.

AI gives that proven rhythm a new interface.

From playing to directing

ChatGPT now reaches more than one billion weekly active users, making conversational AI a mainstream software interface rather than an experimental one.

Exodus-Protocol asks what happens when that interface becomes the primary way to command a civilization.

The player remains the commander, but works together with an AI Assistant that can retrieve game information, explain what has changed, provide strategic advice, and perform permitted actions.

Instead of navigating through layers of menus, the player can communicate an objective:

I need more Graphene. Look at my planets and tell me where investing makes the most sense.

Or:

I want to be ready to colonize another system by the weekend. Work out what we need.

The AI can investigate the current situation, compare options, and help execute the resulting strategy.

Visual information remains visual. When a galaxy map, construction queue, planet overview, or fleet interface communicates something better than words, the Assistant can bring that component into the conversation.

The player's role gradually shifts from operating every individual system to directing a civilization.

But the AI is deliberately not omniscient.

It only knows what the player is allowed to know. If an enemy fleet has only been partially detected, the AI must reason from incomplete intelligence. If a solar system remains undiscovered, it cannot secretly inspect it.

And the language model never determines gameplay reality.

The backend determines resources, travel, combat, progression, and what information is visible. AI understands that reality, explains it, and acts upon it.

AI should remove unnecessary complexity without removing the uncertainty that makes strategy interesting.

A universe built for scale

Exodus-Protocol takes place in a universe that exists independently of the individual player.

The galaxy is procedurally generated from configurable rules. Solar systems, planets, resources, routes, and planetary characteristics exist before somebody discovers them.

Artificial intelligence adds the next layer.

The generated characteristics of a planet provide the basis for its artwork and description. A lore knowledge base containing history, characters, and events provides additional narrative context. That information becomes available to the AI Assistant when the world is eventually discovered.

The simulation determines the world. AI helps turn that world into a place.

This approach also allows us to think differently about scale.

Our background is in building large-scale interactive experiences in which very large audiences participate simultaneously. We are applying what we have learned about concurrency and shared experiences to a persistent game world.

Instead of manually designing a galaxy with a fixed number of locations, generation can respond to the scale required by the game. Our long-term ambition is to investigate persistent worlds capable of supporting populations far beyond traditional multiplayer sessions, potentially reaching hundreds of thousands or even a million concurrent players if demand and infrastructure justify it.

At that scale, procedural generation and automation are not simply creative conveniences. They become part of the architecture required to make the world possible.

A game that creates its own entertainment

A persistent multiplayer world continuously produces events.

Players explore. Colonies grow. Alliances form. Fleets travel. Resources become valuable. Conflicts begin.

Exodus-Protocol is being designed so that the most interesting of those events can become entertainment themselves.

The Always-On Galaxy Broadcast allows the backend to continuously select activity worth showing and create a playlist for a dedicated visualization system. A real-time game engine running in the cloud can transform authoritative game events into cinematic scenes, including camera positioning, movement, effects, and sound.

Quiet periods might show ships traveling through space, distant planets, or industrial activity.

Then a major battle occurs and the broadcast moves there.

The game simulation has already determined what happened. The cinematic system simply determines how people experience it.

The result is something between a game broadcast and slow television: a universe that can run on a second screen while somebody works, occasionally becoming dramatic when something important happens.

It also creates a new distribution opportunity.

Free ad-supported streaming television has become a significant category in its own right. Gracenote tracked 2,172 FAST channels globally in Q3 2026, up 17.5% year over year, while Nielsen previously measured FAST at 6.8% of total U.S. television consumption.

That creates the possibility that Exodus-Protocol is not only discovered through a game store or website.

Someone could encounter the galaxy on YouTube, Twitch, or eventually a FAST channel, watch something created by actual player activity, become curious about what caused it, and eventually enter the universe themselves.

The players create the events. The game turns those events into entertainment.

Reinventing how games are built

The scope of Exodus-Protocol is larger than something we could realistically have attempted with a team of our current size using a traditional development model.

That is intentional.

AI is not only part of the game. It has become a fundamental part of how the game is created.

Our own process evolved through trial and error. We started with AI as an assistant that could generate or autocomplete code. Over time, it became part of specification, architecture, implementation, testing, content creation, art production, and review.

The development process increasingly looks like:

Define → Specify → Challenge → Plan → Implement with agents → Test → Review → Refine

This is very different from simply buying developers an AI subscription.

Google's DORA research, based on nearly 5,000 technology professionals, found AI adoption approaching 90%, but concluded that AI behaves primarily as an amplifier. Strong workflows, clear organizational knowledge, good platforms, testing, and culture determine whether organizations actually benefit.

The game industry itself remains much more conflicted. GDC's 2026 survey found that 36% of industry professionals already use generative AI at work, while 52% believe it is having a negative impact on the industry.

We understand those concerns, but we see a different opportunity.

AI does not remove the need for ideas, architecture, game design, art direction, or judgment. Someone still needs to decide what should exist, whether it is good, and whether it belongs in the game.

What AI can change is how much a small number of experienced people are capable of producing.

For independent studios, that could be significant.

The opportunity is not simply to make the same game with fewer people. It is to allow a small team to attempt a game that would previously have been outside its reach.

Exodus-Protocol is our attempt to demonstrate that in practice.

More than one game

The final part of the thesis goes beyond Exodus-Protocol itself.

The game is being built on top of a reusable persistent-strategy framework.

Many of the difficult systems underneath Exodus-Protocol are not inherently about space: persistence, resources, construction, progression, units, movement, logistics, exploration, intelligence, alliances, combat, and events.

Space is the first thematic implementation of that foundation.

A future game could take place in a fantasy world filled with kingdoms, orcs, and dragons.

Another could be medieval.

Another could replace planets with islands, galactic lanes with sea routes, and starships with naval fleets.

The same foundations could even support a historical war game with nations, factories, armies, and strategic territories.

These would not simply be visual reskins. Each game would require its own mechanics, balancing, creative direction, and identity.

But they would not have to begin from nothing.

Generative AI makes that framework considerably more interesting because large parts of the thematic layer can also be created and adapted faster: artwork, descriptions, lore, technology names, structures, units, and other content can be generated under human direction.

Instead of starting the next game with an empty repository, we want to start with a working world simulation and ask what needs to change.

If that works, much of the expensive foundational work is concentrated in the first game.

The framework, the development process, and the AI tooling created while building Exodus-Protocol can fundamentally change how every game that follows is produced.

The experiment

We do not know yet whether every part of this thesis will work.

That is the reason Exodus-Protocol exists.

Can an AI Assistant become a natural way to command a deep strategy game?

Can a persistent game fit around someone's day instead of demanding hours of uninterrupted attention?

Can a small experienced team use AI to build something with the ambition normally associated with a much larger organization?

Can a game continuously create entertainment from the actions of its players?

And can the framework created for one world make the next world dramatically faster to build?

Some assumptions will survive. Others will not.

We intend to find out by building the game, putting it in front of real players, and sharing what we learn.

Exodus-Protocol is not our prediction of what the future of games will look like. It is our attempt to build one possible version of it.

References

Every external claim on this page, with its source. Listed because a thesis that cites research should let you check it — and because several of these numbers date quickly, so the link matters more than the figure.