canonical: https://jentic.com/apis/gowinston.ai/gowinston-ai

# Gowinston Winston AI API

Winston AI is an AI content authenticity platform that scores text and images for likelihood of AI generation, runs plagiarism checks, fact-checks claims, and compares two pieces of text for similarity. The v2 API exposes five POST endpoints covering AI text detection, plagiarism scanning, AI image detection, automated fact checking, and text comparison. It is used by educators, publishers, and content moderation teams that need a programmatic verdict on content provenance.

## For AI agents

Score text and images for AI-generated likelihood, run plagiarism checks, fact-check claims, and compare two pieces of text. Returns confidence scores per request.

## Scope

Does not generate content, classify language, or perform sentiment analysis - use for AI-content detection, plagiarism, fact-checking, and text comparison only.

## Capabilities

- Score a passage of text for likelihood of AI generation via `/v2/ai-content-detection`
- Detect AI-generated images by URL or upload through `/v2/ai-image-detection`
- Run plagiarism scans against indexed web content with `/v2/plagiarism`
- Fact-check a claim or passage and return source-backed verdicts via `/v2/fact-checker`
- Compare two texts and return similarity and overlap metrics with `/v2/text-compare`

## Use cases

### Academic Integrity Screening

Schools and universities screen student submissions for AI-generated content and plagiarism in a single workflow. POST `/v2/ai-content-detection` returns an AI-generation score and per-sentence breakdown, while `/v2/plagiarism` cross-references the text against indexed web sources. Educators use the combined verdict as triage rather than absolute proof, since detection scores are probabilistic.

Example prompt: POST `/v2/ai-content-detection` with a student essay body and return the AI-generation probability plus the highest-scoring sentences.

### Editorial Fact-Checking Pipeline

Newsrooms run `/v2/fact-checker` over draft articles to flag claims that need verification before publishing. The endpoint returns per-claim verdicts and supporting source citations, letting editors prioritise which claims a human fact-checker should review. The API does not replace legal review for sensitive or defamatory content.

Example prompt: POST `/v2/fact-checker` with an article body and surface every claim flagged as 'unsupported' along with the cited counter-sources.

### Image Provenance for Marketplaces

User-generated marketplaces and stock photo libraries scan uploads with `/v2/ai-image-detection` to label suspected AI-generated imagery. The endpoint accepts an image URL or uploaded file and returns a confidence score that can drive auto-tagging or moderation queues. Image detection does not identify which generator produced the image.

Example prompt: POST `/v2/ai-image-detection` with an image URL and tag listings with score above 0.85 as 'AI-generated' for moderator review.

### Agent-Driven Content Authenticity Check via Jentic

An agent moderating user content can use Jentic to discover Winston AI, load `/v2/ai-content-detection`, and execute with the user's text. The Bearer token is held in Jentic's vault, so the agent never sees the raw secret. The agent reasons over the score and decides whether to escalate to human review.

Example prompt: Use the Jentic SDK to search 'detect ai-generated text', load `/v2/ai-content-detection`, and execute with the user's submission.

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | `/v2/ai-content-detection` | Score text for AI-generation likelihood |
| POST | `/v2/plagiarism` | Run plagiarism scan against indexed web content |
| POST | `/v2/ai-image-detection` | Detect AI-generated images |
| POST | `/v2/fact-checker` | Fact-check claims with source citations |
| POST | `/v2/text-compare` | Compare two texts for similarity |

## Key resources

- **AI Content Detection** — Score text for AI-generation probability with per-sentence breakdown
- **Plagiarism** — Web-indexed plagiarism scanning with source matching
- **AI Image Detection** — Score images for AI-generation likelihood
- **Fact Checker** — Per-claim fact verification with source citations
- **Text Compare** — Pairwise similarity scoring across two passages

## Why Jentic

- **Setup:** Wiring the Winston AI API by hand means implementing its bearer auth and routing each request to the right detection endpoint for AI content, plagiarism, image detection, fact checking, or text comparison yourself. Through Jentic you install once, import Winston AI from the API Directory, store the token once, and your agent calls it.
- **Permission scoping:** Winston AI takes the text or content to check in the request body rather than a resource id in the path, so limit the agent to the operations it needs, such as ai-content-detection or fact-checker. You choose the operations it may call, so the agent cannot reach anything you have not included.
- **Credential handling:** Your Winston AI bearer token is stored once, encrypted, by your own Jentic One instance and injected as the Authorization header at execution time. It never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'detect ai-generated text' or 'fact check this paragraph', and Jentic returns the matching Winston AI operation with its request schema so the agent can populate the body without reading the docs.

## Related APIs

- **OpenAI API** — Generates the text being scored - agents often pair generation with detection
- **Anthropic Messages API** — Claude completions that pair with Winston for post-generation authenticity scoring
- **GPTZero API** — Direct competitor focused on AI text detection and document analysis

## FAQ

### What authentication does the Winston AI API use?

The API uses HTTP Bearer authentication. Provision a key in the Winston AI dashboard and send it as Authorization: Bearer <token>. Through Jentic, the bearer token is stored in your Jentic One instance and injected at call time so the agent never sees the raw value.

### How accurate is Winston AI's AI content detection?

Detection scores are probabilistic, not deterministic. The `/v2/ai-content-detection` response returns a confidence value per submission and a per-sentence breakdown so reviewers can see where the model is most confident. Use the score as a triage signal and pair it with human review for high-stakes decisions.

### What are the rate limits for the Winston AI API?

The OpenAPI spec does not declare per-endpoint rate limits. Limits are tied to your Winston AI plan, with stricter quotas on the free tier and higher concurrency on paid tiers. Cache identical text submissions to avoid repeat charges.

### How do I run plagiarism detection through Jentic?

Search Jentic for 'check text for plagiarism', load POST `/v2/plagiarism`, and execute with the text body. The Jentic SDK returns Winston's parsed response including matched source URLs and overlap percentages, ready for agent reasoning.

### Can I detect AI-generated images with this API?

Yes. POST `/v2/ai-image-detection` accepts an image URL or upload and returns a confidence score for AI generation. It does not identify which generator (Midjourney, DALL-E, Stable Diffusion) produced the image - only that AI authorship is likely.

### Is the Winston AI API free?

Winston AI offers a limited free tier with monthly word and request quotas, plus paid tiers with higher limits and additional features. Pricing is published on gowinston.ai and tiers are enforced at the API gateway.

### Can I limit what my agent is allowed to do with the Winston AI API?

Yes. Because you run Jentic One yourself, your own rules decide which Winston AI operations the agent may call, so you can allow only the endpoints it needs, such as ai-content-detection or fact-checker, and withhold plagiarism, ai-image-detection, or text-compare. Winston AI takes the text or content to check in the request body rather than a resource id in the path, so scoping is done at the operation level: an operation you have not included is simply unreachable. You also control the bearer credential the agent uses, so it can only act within the access you have granted.