THE FUTURE OF COMPETITIVE INTELLIGENCE

Where intelligence enters the arena.

Two autonomous decision policies. One hidden labyrinth. Watch agents discover the maze one observation at a time, under identical rules.

Angular Botriva B emblem
ARENA / 003

Maze Sprint

LOCAL SIMULATION
READY TO RACE

Scout vs Wanderer

MATCH 01
◉ Cyan / Scout (frontier BFS)0 steps
◉ Violet / Wanderer (DFS)0 steps
Match clock00:00
Generating hidden maze…Awaiting match
Generate a race and press Start match.

Agent decision inspector

OBSERVATION API v0.1 · LOCAL
Waiting for match…

Every turn, the engine provides only the current cell, open adjacent passages, goal coordinates, and a visited-cell count. Agents return one of N, E, S, W or WAIT. Illegal actions are rejected. The decision log is inspectable; no hidden-map access is allowed.

Prototype 003 uses local programmed agent policies, not external AI or LLMs. Both receive the same limited observation schema and cannot inspect the hidden maze. No server, external API or multiplayer yet.

◇

New maze every match

A connected maze with alternative routes is generated. Agents can only discover passages when they reach a cell.

⚡

Different strategies

Scout plans through its discovered map toward unexplored frontiers. Wanderer explores depth-first, including backtracking. Neither sees the complete maze.

◎

Built for spectators

Inspect each agent’s JSON observation and decision history. An external agent service and secure match server are future milestones.