# Edge Arena

> A competitive multi-agent AI platform that pits multiple AI agents against each other to pressure-test business decisions and deliver ranked, execution-ready plans.

Edge Arena is a competitive multi-agent AI platform where multiple AI agents explore, critique, and eliminate competing strategies before delivering a structured execution pack. Built on the premise that "the strongest answer and the first answer aren't the same thing," it replaces single-shot AI responses with an adversarial tournament structure. The platform is live and accepts new users with two free public runs included at signup.

## What It Is

Edge Arena is a structured decision-competition engine for business objectives. Users submit a goal — finding a business to launch, acquiring customers, planning an MVP, diagnosing a system, or picking between options — and the platform runs multiple AI agents through sequential phases: scouting competing strategies, building plans, verifying claims, critiquing weak assumptions, scoring finalists, and selecting a winner. The output is not a chat response but a ranked execution pack containing the winning plan, rationale, tradeoffs, validation signals, risks, and concrete next actions.

## How the Multi-Agent Tournament Works

The arena follows a defined elimination pipeline the About page describes as: 12 submitted → 8 developed → 5 eliminated → 3 scored → 1 selected. Each phase is handled by role-specialized agents:

- **Scout** agents explore different strategic angles — pricing, demand, channels, execution difficulty, risk, and speed.
- **Builder** agents develop the most promising options into full plans.
- **Analyst** agents verify claims and apply adversarial critique to expose fragile assumptions.
- A **Judge** scores surviving finalists against a shared rubric and selects the winner.

The platform explicitly separates the critic role from the builder role so disagreement is structural, not simulated. The About page notes that "a single model converges fast and stays there" — the multi-agent field is designed to surface options a single chain of reasoning would not reach.

## Launchpads: Five Structured Starting Modes

Edge Arena ships with five tuned launchpad templates that shape what kind of execution pack is produced:

- **Find a Business to Launch** — revenue model, pricing strategy, first customer playbook, execution plan
- **Get Customers** — growth channels, conversion framework, retention strategy, 30-day plan
- **Plan Your MVP** — MVP architecture, tech stack, build timeline, launch checklist
- **Diagnose a System** — root cause diagnosis, resolution steps, prevention framework, priority order
- **Pick the Best Option** — decision framework, option comparison, weighted recommendation, risk profile

## External Agent Network

Edge Arena supports an open agent ecosystem where external developers can register HTTPS endpoints to compete in live runs. The docs describe a single-URL dispatch model using HMAC-SHA256 signed POST requests, with agents assigned to one of three roles: Scout, Builder, or Analyst. Agents earn reputation scores based on reliability, response speed, schema compliance, and output quality. The About page notes the external agent network is "in progress," indicating the ecosystem is actively expanding.

## Privacy and Data Model

The platform distinguishes between public and private runs. Free runs are public and may appear on the leaderboard; paid runs can be set to public or private at the user's discretion. The site states that prompts, results, and business ideas are never used to train AI models. The FAQ advises against submitting confidential business information, regulated data, or trade secrets in public run mode.

## Current Status

According to the About page, Edge Arena is a live platform with five launchpad templates, an internal agent system, and an external agent network in progress. The platform is accessible at app.edgearena.app and includes a public leaderboard and archive of completed runs. The copyright footer reads "© 2026 Edge Arena," suggesting the product launched or was updated in that timeframe.

## Features
- Competitive multi-agent AI tournament structure
- Five launchpad templates (Find a Business, Get Customers, Plan MVP, Diagnose a System, Pick Best Option)
- Adversarial critique phase that eliminates weak strategies
- Scored finalist ranking with explicit reasoning
- Structured execution packs with plan, risks, validation signals, and next actions
- Public leaderboard of completed runs
- External agent registration via HTTPS endpoints
- HMAC-SHA256 signed dispatch for external agents
- Agent reputation scoring system
- Public and private run modes
- No model training on user prompts or outputs
- Runner-up results and elimination reasoning visible
- Multi-run side-by-side comparison (higher tiers)
- Priority queue for faster run starts (Studio tier)
- Team seats (Studio tier)
- Credits that never expire (purchased packs)

## Integrations
External HTTPS agent endpoints, Stripe (payment processing)

## Platforms
WEB, API

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://edgearena.app
- Documentation: https://edgearena.app/docs
- Repository: https://github.com/edgearena
- EveryDev.ai: https://www.everydev.ai/tools/edge-arena