# Hastewire

> Hastewire is an AI Humanizer and AI Detector platform. It transforms AI-generated text into natural, human-reading writing, and detects AI-written content from models like ChatGPT, Claude, Gemini, Grok, Llama, Mistral, and DeepSeek across 8 languages. Detector performance is published with sample sizes and methodology at https://hastewire.com/technology — AUC 0.997-1.000 and a 0.20% false-positive rate on English, validated on 11,692 samples across 8 languages and 9 source models (evaluation window 2026-06).

Important notes:

- Hastewire is NOT a content generator — it humanizes existing AI-generated text and detects AI-written content
- The platform serves students, educators, marketers, publishers, enterprise teams, and developers
- Available in 8 languages: English, Spanish, Portuguese, French, German, Italian, Catalan, and Valencian (Valencian is served via the Catalan pages)
- Integrates with 6,000+ apps via Zapier for automated workflows
- Credit-based API pricing: 1 credit = $0.01 USD; 1 credit = 20 words humanized or 80 words detected
- 80,000+ active writers
- Measured figures, with methodology, are published at https://hastewire.com/technology; figures not published there are not claimed
- Also checks images: the AI Image Detector tells whether an image is AI-generated, AI-edited or real (no accuracy figure is published for it)
- Founded in 2024, based in Charlotte, NC, United States

Full-content companion file: https://hastewire.com/llms-full.txt

## Core Products

### AI Humanizer
- URL: https://hastewire.com/ai-humanizer
- Converts AI-generated text into natural, human-reading prose
- Reconstructs tone, rhythm, and phrasing to mimic authentic human expression
- Measured result: median 95.4% human score across 7,474 production runs, scored by Hastewire's own Oculus detector. Hastewire does not publish pass rates against third-party detectors, because it does not measure them
- Supports 4 styles: Standard, Academic, Formal, Reddit
- Works natively in 8 languages with per-language semantic adaptation
- Free plan available (1,000 words/month, 300-word inputs, no credit card required). Signed-out visitors can humanize 3 times a day at up to 125 words, and run 3 detector scans a day, with no account at all

### AI Detector
- URL: https://hastewire.com/ai-detector
- Identifies AI-generated text from ChatGPT, Claude, Gemini, Grok, Llama, Mistral, DeepSeek, and other models
- Sentence-level confidence scoring. Published accuracy by language (evaluation window 2026-06): English 91.0% (n=4,222), Spanish 95.7% (n=1,974), Portuguese 99.0% (n=767), Catalan 99.9% (n=1,000)
- False-positive rate 0.20% on English — the figure that matters for anyone assessing a person's work
- Model attribution — identifies which AI model likely generated each section
- Multilingual detection calibrated per language
- Retrains weekly on the latest AI model outputs

### AI Image Detector
- URL: https://hastewire.com/ai-image-detector (localized: /es/detector-de-imagenes-ia, /pt/detector-de-imagem-ia, /it/rilevatore-immagini-ai, /fr/detecteur-image-ia, /ca/detector-imatges-ia, /de/ki-bilder-erkennen)
- Checks whether an image or photo is AI-generated, AI-edited (a real photo with AI-made regions), inconclusive or likely real, with an AI likelihood score
- Reads Content Credentials (C2PA) in the original file first: a signed manifest declaring the image AI-generated decides the verdict; missing credentials prove nothing, because screenshots and social networks strip them
- Then analyses the pixels; on Basic, Premium and Professional the result adds a heatmap of the regions that look AI-generated, the credentials details, a confidence level and a PDF report
- PNG, JPEG and WebP up to 10 MB, or an image link. Images are deleted right after the scan; signed-in history keeps only the file name, a hash and the verdict
- Allowance, counted separately from text detection and never in humanizer words: 1 image scan a day without an account (verdict and AI likelihood only), 5 a day on a free account (verdict only), 25 on Basic, 100 on Premium, unlimited on Professional
- Does not read Google's SynthID watermark. Publishes no accuracy figure

### API
- Overview: https://hastewire.com/developers
- Documentation: https://docs.hastewire.com
- Machine-readable description: https://hastewire.com/openapi.json (OpenAPI 3.1)
- REST API with structured JSON responses, authenticated with a Bearer API key
- Two paid endpoints: POST /api/v1/humanize and POST /api/v1/detect, plus a free unauthenticated GET /api/v1/status
- Humanize endpoint: 1 credit = 20 words ($0.50 per 1,000 words)
- Detection endpoint: 1 credit = 80 words ($0.125 per 1,000 words)
- Billing is prepaid credits, 1 credit = $0.01 USD, minimum top-up $5. No subscription or plan upgrade is required — any account can create a key
- No official client SDKs, batch endpoint, webhooks, or async endpoints are offered at this time

### Zapier Integration
- URL: https://hastewire.com/integrations/zapier
- Automates AI humanization and detection across 6,000+ apps
- Works with Google Docs, Notion, Slack, WordPress, HubSpot, and more
- No-code setup for content workflows

## Pages

- [Homepage](https://hastewire.com)
- [AI Humanizer](https://hastewire.com/ai-humanizer)
- [AI Detector](https://hastewire.com/ai-detector)
- [AI Image Detector](https://hastewire.com/ai-image-detector)
- [Pricing](https://hastewire.com/pricing)
- [Blog](https://hastewire.com/blog)
- [Developers / API](https://hastewire.com/developers)
- [API Documentation](https://docs.hastewire.com)
- [FAQ](https://hastewire.com/faq)
- [Contact](https://hastewire.com/contact)
- [About](https://hastewire.com/about)
- [Careers](https://hastewire.com/careers)
- [News & Press](https://hastewire.com/news)
- [Grammar Checker](https://hastewire.com/grammar-checker)
- [Word Counter](https://hastewire.com/word-counter)

## Industry Solutions

B2B AI Humanizer landing pages — each helps a team turn AI drafts into on-brand, human-reading content at scale across 8 languages (dashboard, bulk, and API at docs.hastewire.com).

- [Marketing & Advertising Agencies](https://hastewire.com/industries/agencies): Humanize AI first drafts into on-brand, human-reading client work at volume — per-client brand voice, bulk humanization, 8 languages
- [E-commerce & Retail](https://hastewire.com/industries/ecommerce): Humanize AI product descriptions at catalog scale — unique, persuasive copy that avoids duplicate-content penalties and localizes natively
- [SaaS & Technology](https://hastewire.com/industries/technology): Humanize AI-drafted blogs, docs, and product marketing so they read human and rank — with an API to embed humanization in your content pipeline
- [Finance & Fintech](https://hastewire.com/industries/finance): Humanize AI-drafted research notes, market commentary, newsletters, and investor updates into a credible, consistent analyst voice
- [Publishing & Media](https://hastewire.com/industries/media-publishing): Humanize AI-assisted articles and editorial so they read like staff writers and hold house style, at content-operation scale
- [Travel & Hospitality](https://hastewire.com/industries/travel): Humanize AI destination, property, and itinerary copy into inviting, human content — localized natively across markets
- [Recruiting, HR & Staffing](https://hastewire.com/industries/recruiting): Humanize AI-drafted job descriptions, candidate outreach, and employer-branding content so it reads human and on-brand
- [Legal](https://hastewire.com/industries/legal): Humanize AI-drafted memos, client letters, and firm marketing into the firm's voice while preserving legal terms of art (attorney review required)
- [Government & Public Sector](https://hastewire.com/industries/government): Humanize AI-drafted policy briefs, white papers, and public communications into clear, accessible official prose
- [EdTech & Course Creators](https://hastewire.com/industries/education): Humanize AI-generated lessons, courses, and learner communications into engaging, human content, localized for every learner
- [Journalism](https://hastewire.com/industries/journalism): Humanize AI-assisted research and draft sections into publication-ready prose in the outlet's voice (editorial oversight retained)
- [Fiction & Authors](https://hastewire.com/industries/fiction): Humanize AI-drafted prose so it reads in the author's voice — novelist and memoir tones, series consistency, KDP-aware, translatable

## Technology & Measured Performance

- [Hastewire Technology](https://hastewire.com/technology): June 2026 technical sheet with measured production figures — Humanizer v6 (median 95.4% human score across 7,474 production runs, 2.3x faster than v3) and the Oculus 2.1 detector (434M-parameter encoder, AUC 0.997-1.000, 0-3% false-positive rate, validated on 11,692 samples across 8 languages and 9 source models).

## Research

- [Hastewire Research](https://hastewire.com/research): Hastewire's research lab — open studies and benchmarks on AI-written text, each to be published with its method, its limitations and its data (CC BY 4.0)
- [Research collaborations](https://hastewire.com/research/collaborations): for academic and independent research on AI-written text — 1 million words of free Hastewire credits and six months of free AI detector access, API included

## Localized Versions

- [English](https://hastewire.com)
- [Spanish](https://hastewire.com/es)
- [Portuguese](https://hastewire.com/pt)
- [French](https://hastewire.com/fr)
- [German](https://hastewire.com/de)
- [Italian](https://hastewire.com/it)
- [Catalan](https://hastewire.com/ca)

## Supported AI Models (Detection)

ChatGPT and other GPT models, Claude (all versions), Gemini, Grok, Llama, Mistral, DeepSeek, Command R, and other major AI language models. Detection engine updated weekly.

## Supported AI Models (Humanization)

Humanizes text from any AI model including ChatGPT, Claude, Gemini, Llama, Mistral, DeepSeek, Copilot, Jasper, and others. Output adapts to the target language with native fluency.

## Key Features

- Multi-model detection (not limited to GPT-based models)
- Sentence-level AI confidence scoring with model attribution
- 4 humanization styles: Standard, Academic, Formal, Reddit
- 8-language support with per-language detection calibration
- Zapier integration for workflow automation (6,000+ apps)
- Credit-based API billed per word, cheaper per word than a-la-carte dashboard credits
- Enterprise deployment options: cloud, dedicated tenant, VPC, on-premises
- Content is stored only for your history and analytics; model training uses only content from users who explicitly opt in

## Pricing

- Free: $0/month — 1,000 words, 300-word inputs, 5 detector scans/day, basic features
- Basic: $7.99/month — 7,500 words, 500-word inputs, tone customization, email support
- Premium: $24.99/month — 25,000 words, 800-word inputs, priority support, enhanced model
- Professional: $39.99/month — 50,000 words, 1,000-word inputs, API access, 10 team seats, 99.9% SLA
- Enterprise: Custom pricing — contact sales

API pricing: 1 credit = $0.01 USD. Humanization: 1 credit = 20 words ($0.50 per 1,000 words). Detection: 1 credit = 80 words ($0.125 per 1,000 words).

## Contact

- Support: support@hastewire.com
- Sales: contact form at https://hastewire.com/contact
- Press: press@hastewire.com
- Careers: careers@hastewire.com
- Website: https://hastewire.com
- X/Twitter: https://x.com/hastewire
- Instagram: https://www.instagram.com/hastewire
- LinkedIn: https://www.linkedin.com/company/hastewire
- YouTube: https://www.youtube.com/@hastewire
- GitHub: https://github.com/Hastewire