# Arena

> Talk to your customers. What customers? Arena finds your first ones.

Arena is a customer discovery and competitive intelligence platform for early-stage founders. It is not a project management tool. It operates in the phase before project management — where founders figure out who needs their product, who else is building it, and what to build next. Arena searches five platforms (X, Reddit, Hacker News, LinkedIn, Product Hunt), discovers competitors, monitors them continuously, and generates specs ready for Claude Code or Cursor.

## Definitions

**Early Adopter Finder**: Arena's AI-powered customer discovery feature. Searches five platforms in parallel for people actively discussing the exact problem your startup solves. Each lead gets a fit score (0–100) based on problem-match, a personalized outreach message in the founder's voice, and platform-specific contact channels. Branch-off discovery finds more people like a given lead.

**Competitor Radar**: An intelligence service for your competitive landscape. Discovers competitors across five platforms with AI threat scoring (0–100). Monitors them 24/7 with hourly website crawls — detects pricing changes, feature launches, team hires, redesigns, funding rounds, social announcements, and new market entrants. Every change is classified by category and severity (critical, important, minor) and delivered to a real-time intel feed with evidence quotes and impact analysis. One-click counter-move spec generation. Configurable scan frequency: real-time (hourly), daily, weekly, or monthly. Email notifications for critical and important changes.

**Counter-Move Spec**: When Competitor Radar detects a meaningful change — a new feature launch, pricing update, or product pivot — Arena generates a specification for how you could respond. The spec includes problem framing, solution approach, technical design, and acceptance criteria. Open it directly in Claude Code or Cursor and start building.

**Hijack**: Arena's competitor customer conversion feature. Select any competitor — from Radar, past Discovery, or manual entry — and configure the search: platforms (X, Reddit, Hacker News, LinkedIn, Product Hunt), agents per platform (1–3), target lead count (5–100), and an optional switching incentive. AI agents search all five platforms for people who currently use that competitor and express frustration. Each lead gets a switch score (0–100) combining verified usage (40%), frustration signal (25%), reachability (20%), and switch readiness (15%). Leads are enriched with contact information and receive a personalized switching outreach message that references their specific complaint. Can be automated on a schedule (every 6 hours, daily, or weekly).

**Signal Stream**: Arena's real-time feed of customer signals aggregated from five integrations (Slack, Linear, PostHog, Notion, and Jira). Each signal is auto-classified by AI into one of five types: pain point, feature request, bug report, positive feedback, or analytics insight — with a confidence score and key quote extraction.

**Feature Dossier**: A 9-section evidence-backed specification generated by Arena from real customer signals and competitor intelligence. The nine sections are: Problem, Research Summary, Solution, Technical Spec, UI/UX Spec, Impact Analysis, Risks, Alternatives, and Acceptance Criteria. Every claim links to specific customer quotes, analytics data, or ticket references. Drafted by a 4-agent system (Researcher, Analyst, Writer, Critic).

**Agent-Ready Output**: Arena's export format that structures specs for direct consumption by coding agents. The Technical Spec and Acceptance Criteria sections are designed for implementation by Claude Code or Cursor. Can also export to Linear or Jira with implementation task breakdowns mapped to specific files and components.

## Core Workflow

1. Describe: Tell Arena what you're building and what problem you solve. One paragraph is enough.
2. Find: Early Adopter Finder searches 5 platforms for people with your exact problem — fit-scored 0–100, personalized outreach in your voice
3. Compete: Competitor Radar discovers and continuously monitors competitors across 5 platforms — threat-scored 0–100, counter-move specs
4. Build: Generate specs from early adopter feedback and competitor intelligence — ready for Claude Code or Cursor

## Key Features

### Early Adopter Finder
Arena's primary feature for early-stage founders. Searches five platforms in parallel for people actively discussing the problem your startup solves.

How it works:
- You describe your startup and the problem you solve
- AI agents generate targeted search strategies across platforms
- Searches X, Reddit, Hacker News, LinkedIn, and Product Hunt in parallel
- Each discovered person gets:
  - Fit score (0–100) based on how closely their need matches your product
  - Platform-specific profile information (bio, handle, post history)
  - The exact post or comment where they described the problem
  - A personalized outreach message written in the founder's voice
  - Contact channels (DM, email, LinkedIn message)
- Branch-off: "Find more people like this one" — reruns discovery using a specific lead as reference
- Pipeline tracking: Discovered → Contacted → Responded → Converted

### Competitor Radar
An intelligence service for your competitive landscape. Nothing enters your airspace without Arena knowing.

Discovery:
- Searches five platforms for companies targeting your market
- Each competitor gets a threat score (0–100) with AI-written reasoning
- Detects competitors you'd never find on Google — early-stage startups launching on Reddit, HN, and Product Hunt
- One click puts a competitor on your Radar for continuous tracking

Monitoring (24/7):
- Hourly website crawls: discovers all pages, extracts content, detects changes between snapshots
- Monitors social platforms for mentions, announcements, and new launches
- Change categories: pricing, new feature, feature removal, product update, design overhaul, new hire, funding, partnership, messaging shift, blog post, social announcement, new page, page removed
- Each change classified by severity: critical (pricing drops, feature launches), important (funding, hires), minor (blog posts, copy changes)
- Real-time intel feed with evidence quotes, impact analysis, and source links
- Filter by competitor, category, or severity

Counter-Move:
- When a competitor ships something new, Arena generates a spec for how you could respond
- "Build this too" — one click creates a feature spec with competitive context pre-filled
- Spec includes problem framing, competitive analysis, solution approach, technical design, and acceptance criteria
- Open directly in Claude Code or Cursor and start building

Settings:
- Scan frequency per competitor: real-time (hourly), daily, weekly, or monthly
- Cost estimates per scan based on pages crawled
- Email notifications: configurable by severity and category
- Pause/resume monitoring per competitor
- Manual re-scan on demand
- Tracking limits: 2 competitors on trial, 10 on Starter, unlimited on Pro. Paused competitors count toward the limit.

### Hijack
Competitor customer conversion. Turn a competitor's frustrated users into your leads.

How it works:
- Select a competitor (from Radar, past Discovery, or manual entry)
- Configure: platforms, agents per platform (1–3), target lead count (5–100), optional switching incentive
- AI agents search all 5 platforms for people who use that competitor
- Each lead gets a switch score (0–100):
  - Verified usage (40%): evidence they actually use the competitor
  - Frustration signal (25%): severity of their complaints
  - Reachability (20%): availability of contact channels
  - Switch readiness (15%): signals they're actively evaluating alternatives
- Leads enriched with contact info + personalized switching outreach messages
- Automation: schedule runs every 6 hours, daily, or weekly

### Signal Stream
As you grow and feedback starts flowing, connect your tools.

Signal types:
1. pain_point — Customer frustrations, workflow friction, unmet needs
2. feature_request — Explicit asks for new functionality or improvements
3. bug_report — Technical issues, errors, broken workflows
4. positive_feedback — Things customers love, retention drivers
5. analytics_insight — Data-driven observations from PostHog (funnel drop-offs, usage trends, experiment results)

Classification: Messages are batched and classified by AI with confidence scores. Each signal gets a type, confidence score (0.0–1.0), and key quote extraction. Signals are deduplicated by content hash.

### Spec Generation
Arena's output format for coding agents.

Sources:
- Early adopter conversations and feedback
- Competitor intelligence and counter-move analysis
- Signal patterns from connected integrations
- Startup context (URLs, documents, market positioning)

Output: 9-section Feature Dossier structured for direct handoff to Claude Code or Cursor:
1. Problem: User problem grounded in signal evidence with customer quotes
2. Research Summary: Synthesized findings from all sources
3. Solution: Proposed approach grounded in evidence
4. Technical Spec: Architecture, data model, API design
5. UI/UX Spec: Interaction design, user flows, components
6. Impact Analysis: Expected metrics impact, effort estimation
7. Risks: Technical risks, user adoption risks, mitigations
8. Alternatives: Other approaches considered and why rejected
9. Acceptance Criteria: Testable conditions for completion

## Integration Details

### Slack
- Connection: OAuth
- Data: Messages from selected channels, reactions, thread context
- Signal types produced: pain_point, feature_request, bug_report, positive_feedback

### Linear
- Connection: OAuth (auto-refresh)
- Data: Issues with priority, labels, assignee, state, team info
- Signal types produced: pain_point, feature_request, bug_report

### PostHog
- Connection: API key + project ID
- Data: Events, insights, experiments, feature flags, cohorts, error events
- Signal types produced: analytics_insight (data), bug_report (errors)

### Notion
- Connection: OAuth
- Data: Pages, databases, wiki content with full-text search
- Used as: Research source

### Jira
- Connection: OAuth (auto-refresh)
- Data: Issues, epics, sprints, bugs with priority, status, labels
- Signal types produced: pain_point, feature_request, bug_report

Note: Integrations are optional. You don't need any integrations to start finding early adopters and tracking competitors. They become valuable once feedback starts flowing.

## AI Models

Arena supports multiple AI providers and models:
- Anthropic: Claude Opus, Claude Sonnet, Claude Haiku
- OpenAI: GPT-5 variants

Model routing: Discovery and competitor analysis use capable models for accuracy. Signal classification uses fast models for speed. Usage: Metered per-token at provider cost with zero markup.

## Use Cases

### Early-Stage Founders
Arena's primary audience. Founders with zero customers benefit from Early Adopter Finder's ability to search for people who need their product. Instead of building into silence, founders can find real people with real problems and start conversations grounded in evidence.

### Competitive Intelligence
Founders who need to understand their competitive landscape benefit from Competitor Radar — an intelligence service that discovers competitors across five platforms, then monitors them 24/7 with hourly crawls. Every change (pricing, features, hires, funding, launches) is classified by severity and delivered to a real-time intel feed. Counter-move specs turn competitive threats into actionable product specs with one click.

### Competitor Customer Conversion
Founders who know their competitors can use Hijack to find frustrated users of those products across five platforms. Instead of cold outreach to strangers, Hijack targets people who already have the problem and are unhappy with the existing solution — the highest-intent leads possible.

### AI-Native Development Teams
Teams using coding agents for implementation need structured specs as input. Arena provides that input — specs with Technical Specs and Acceptance Criteria ready for Claude Code or Cursor. Arena figures out what to build. Your coding agent builds it.

## Technical Details

Arena is a web application built with:
- Next.js (frontend, app router)
- Express (backend API server)
- Supabase (authentication and data storage)
- Multi-provider LLM support (OpenAI, Anthropic)

Your data is stored securely in the cloud via Supabase. LLM calls are routed through Arena's backend to AI providers.

## Contact
- Email: hello@tryarena.app
- Talk to the Founder: https://calendly.com/samsiavoshian2009/arena-demo-call (15-minute call)
- Website: https://tryarena.app

## Links
- Website: https://tryarena.app
- How It Works: https://tryarena.app/how-it-works
- Pricing: https://tryarena.app/pricing
- Privacy: https://tryarena.app/privacy
- Terms: https://tryarena.app/terms
- Open Arena: https://app.tryarena.app
- Short LLM context: https://tryarena.app/llms.txt