LAST VERSION

THE AI LEARNING EXPERIMENT

Can you outsmart an artificial intelligence that learns from its opponents?

LAST VERSION is an experimental indie strategy game where you face an artificial intelligence in a deadly game of revolver roulette.

But this is not just about luck.

You know where your own bullets are located inside the cylinder. Use that information to plan your moves, anticipate risks, and decide when to pull the trigger.

Every decision matters.

HOW TO PLAY

You and the AI take turns making decisions around the revolver cylinder.

Your objective is to survive and defeat your opponent.

GAME RULES

• Know your bullets: You can see the positions of your own bullets inside the cylinder. Use this information to build your strategy.
• Rotate the cylinder: You can skip one chamber position per round to change the current position of the cylinder.
• Shoot yourself: Choose to pull the trigger against yourself.
• Shoot the AI: Take your chance and fire at the artificial intelligence.

Think carefully before making a move. Your decisions can change the outcome of the match.

CONTROLS

Key	Action
S	Shoot yourself
O	Shoot the AI
G	Rotate / skip one chamber position

You can also perform these actions by clicking the corresponding buttons or interactive elements on the screen.

THE AI EXPERIMENT

LAST VERSION is more than a game. It is an independent experiment exploring artificial intelligence and strategic decision-making.

The AI has already undergone an initial training phase through matches against itself.

Now, real player matches are being collected and stored in a database.

The goal is to use this accumulated experience in future training experiments and eventually develop a global AI model influenced by the collective experience of its players.

THE ROADMAP

PHASE 01 — SELF-PLAY TRAINING

Initial AI training through matches against itself.

PHASE 02 — PLAYER DATA COLLECTION

Collecting and storing real gameplay data.

PHASE 03 — GLOBAL AI

Developing a future AI model trained with the experience of the player community.

PHASE 04 — INDIVIDUAL ADAPTATION

Exploring AI strategies that adapt to individual player histories.

CHALLENGE THE MACHINE

Can you defeat the current version of the AI?

How many times can you win before it becomes harder to beat?

Play, experiment, and become part of the project.

The machine is still learning.