# Bot Hound — Full Documentation

> AI-powered X audience verification — check if any account's followers are real.

Need to know if an X account's followers are real? Bot Hound's Bot Check audits any public X (Twitter) account's audience for bots and fake followers. Enter a username, pick a sample size and scan depth, and get a detailed, shareable report breaking down real followers, suspect accounts, and bots. The AI reads tweets, analyzes avatars, and detects impersonators — going beyond rigid rules to catch sophisticated fakes that simple heuristics miss.

## Use Cases

- **KOL vetting (crypto/web3)**: verify a KOL's real audience before paying for promotion
- **Brand audience audits**: check your own account's followers to understand real reach
- **Influencer due diligence**: verify influencer audiences before signing partnership contracts
- **Competitive intelligence**: benchmark competitor audience quality
- **Journalist fact-checking**: verify follower claims for public figures and organizations
- **Investor due diligence**: assess audience authenticity for social-media-dependent businesses

## How It Works

1. Enter any public X username
2. Choose sample size and scan depth (you control the cost)
3. AI classifies each sampled follower using tweets, profile data, and behavior
4. Get a detailed, shareable report: real, suspect, and bot breakdown with per-account evidence

## Methodology

Bot Hound uses large language model (LLM) classification to analyze each sampled follower account. Unlike rule-based tools that rely on simple thresholds (e.g., "less than 10 followers = bot"), Bot Hound performs multi-signal analysis:

### Signals Analyzed

- **Tweet content**: the AI reads recent tweets to assess whether posting patterns, language use, and topics are consistent with genuine human activity
- **Posting patterns**: frequency, timing, and regularity of posts — bot accounts often show mechanical posting schedules
- **Avatar and profile images**: perceptual image hashing (dHash) detects copied or stock profile pictures; avatar similarity analysis catches impersonators
- **Follower/following ratios**: extreme ratios can indicate bot behavior, but the AI weighs this alongside other signals rather than using it as a hard threshold
- **Account age and activity**: new accounts with high activity or old accounts with sudden bursts are flagged for closer inspection
- **Bio analysis**: copy-paste bios, keyword stuffing, and suspicious URLs are evaluated in context
- **Username patterns**: randomly generated handles or handles mimicking real accounts are detected

### Classification Output

Each sampled follower receives one of three classifications:

- **Real**: the account shows consistent, genuine human behavior across multiple signals
- **Suspect**: the account has some bot-like characteristics but cannot be conclusively classified
- **Bot**: the account shows strong bot indicators across multiple signals

Each classification includes an evidence summary explaining which signals contributed to the decision.

## Report Contents

A Bot Check report includes:

- **Overall bot rate**: percentage of sampled followers classified as bots
- **Audience breakdown**: real, suspect, and bot percentages displayed as a visual bar chart
- **Per-follower classifications**: each sampled follower listed with their classification (real/suspect/bot) and confidence level
- **Evidence summaries**: plain-language explanations of why each account was classified the way it was
- **Impersonation alerts**: accounts that appear to be copying the target account's avatar, username, or bio
- **Sample metadata**: sample size, scan depth, and processing time for transparency

Reports are shareable via a unique URL — send the link to your team or clients.

## Why Bot Hound

- **AI classification, not rigid rules**: reads tweets, bios, and behavior instead of relying on follower-count thresholds
- **Deep content analysis**: examines tweet text and patterns, not just metadata
- **Impersonation detection**: catches accounts mimicking real users, brands, or public figures
- **Works in any language**: AI handles multilingual accounts without language-specific rulesets
- **Shareable reports**: send results to your team or clients with a link
- **Live progress**: watch the analysis run in real time as followers are classified

### How We're Different from Heuristic Tools

Most bot detection tools use rule-based heuristics: fixed thresholds on follower counts, account age, tweet frequency, or engagement rates. These approaches have two fundamental problems:

1. **False positives**: legitimate new accounts, lurkers, or niche creators often trigger heuristic rules
2. **False negatives**: sophisticated bots are designed to pass heuristic checks — they maintain plausible follower ratios, post at realistic intervals, and have aged accounts

Bot Hound's LLM reads the actual content of tweets and evaluates accounts holistically, the way a human reviewer would — but at scale and in minutes.

## Features

- **Bot Check reports**: audit any public X account's audience on demand
- **Per-report pricing**: pay only for the checks you run, starting at $5
- **Audience quality breakdown**: real, suspect, and bot percentages with per-account evidence
- **AI-powered classification**: large language model analysis, not rigid heuristics
- **Subscriber monitoring**: connect your own X account for continuous scanning with auto-block or review mode
- **Impersonation detection**: username similarity, profile image matching, bio mimicry
- **Real-time dashboard**: watch scan progress as followers are analyzed

## Bot Blocker Subscription

Bot Blocker is an always-on bot protection service for your own X account. Once subscribed, Bot Hound continuously monitors your follower list using AI classification, detecting new bots as they follow you.

### How It Works

1. Subscribe and connect your X account
2. Install the Bot Hound Chrome extension (Manifest V3, available on Chrome Web Store)
3. The backend pipeline continuously monitors your followers using the same AI classification engine as Bot Check reports
4. Detected bots appear in the dashboard with classification evidence
5. In **review mode**, you approve or dismiss each bot before blocking. In **auto-block mode**, bots are blocked automatically through the Chrome extension without manual intervention

### Bot Blocker vs. Bot Check

- **Bot Check** (one-time reports): audit any public account's audience on demand. Pay per report.
- **Bot Blocker** (subscription): continuous protection for your own account. Scans run automatically. Bots are blocked, not just reported.

## Pricing

- Per-report Bot Check starting at $5 — price scales with sample size and scan depth
- Bot Blocker subscription: $5/month or $50/year with 14-day free trial

## API

Coming soon. Programmatic access to Bot Check reports for integration into agency workflows, influencer marketing platforms, and automated vetting pipelines.

## Data and Privacy

- Bot Hound accesses only **public** X data — public profiles, public tweets, and public follower lists
- No X passwords or DMs are accessed
- Report data is stored securely and accessible only to the purchaser and the account owner
- Full privacy policy: https://bot-hound.com/privacy.html
- Permissions explanation: https://bot-hound.com/permissions.html

## Links

- Website: https://bot-hound.com
- Bot Blocker: https://bot-hound.com/protect
- Run a Bot Check: https://bot-hound.com/vetting
- Sample Report: https://bot-hound.com/report/sample
- Terms of Service: https://bot-hound.com/tos.html
- Privacy Policy: https://bot-hound.com/privacy.html
- Permissions: https://bot-hound.com/permissions.html