Intelligence players can read
The goal is not to hide complexity behind the AI label. It is to build decisions with context, feedback and tools so design can observe, tune and explain behavior—including uncertainty and fallback states.
Three scales of intelligence
AIRI AI explores conversation and persistent memories across sessions. SLEIGHT uses an NPC crowd as a camouflage and deduction system. Water Battle Arena places CPU opponents inside a local arena with four characters and clear objectives. Each case needs a different kind of intelligence, but all of them depend on state that remains inspectable.
How I work
I separate perception, memory, decision and action; define what the player should infer; and expose diagnostic points before increasing sophistication. AI and Multiplayer in Production expands this method through authority, uncertainty and observability in connected systems.
