# SigWise: scam, trust and buyer-intent signals API

> SigWise reads the events and messages on your platform and answers the questions you care about, like is this a scammer, how far can we trust them, and are they ready to buy, as typed signals you can act on in real time.

Website: https://sigwise.ai · Docs: https://sigwise.ai/docs · Docs index for LLMs: https://sigwise.ai/llms.txt

## How it works

1. **Send what happens.** Post events and messages about any object (a user, a listing, an order) to `POST /v1/objects/{object_id}/events`. Ingestion returns `202` in milliseconds, and a burst of activity is folded into a single analysis.
2. **Ask your questions.** Define each signal in plain language with `PUT /v1/signals/{key}`: a yes/no probability (`noul`), a position on a labelled scale (`score`) or one of several categories (`choice`). No model to train.
3. **Act on the answers.** Read typed answers from `GET /v1/objects/{object_id}`, search objects by them (`is_scammer >= 90`), receive signed webhooks, or define rules that alert Slack, email or a webhook when a condition becomes true.

Every new account starts with three signals: `is_scammer`, `trust_score` and `buyer_intent`.

## Two speeds

- **Background (default).** Events are analyzed asynchronously a few seconds after activity settles, over the object's recent events plus a rolling summary of its history.
- **Inline.** Add `"wait": true` to score content before it is published (a message, a listing, a payout change) and get the verdict in the same request.

```json
POST /v1/objects/user-42/events
{ "wait": true, "signals": ["is_scammer"],
  "events": [{ "type": "message", "content": "pay me by wire and I double it" }] }

200 OK
{ "object_id": "user-42", "analyzed": true,
  "answers": [{ "signal": "is_scammer", "type": "noul", "noul": 0.97 }] }
```

## Pricing

No subscription, seats or minimums. Every new account gets $5 of free credit; after that you top up a prepaid balance by card. Each analysis is billed at its actual model cost, typically a fraction of a cent (a short marketplace conversation costs about $0.0001). Usage is broken down by day, model, signal and source, and analyses pause when the balance runs out.

## Use cases

- [Marketplace scam detection](https://sigwise.ai/docs/use-cases/marketplace-scam-detection.md)
- [User trust scores](https://sigwise.ai/docs/use-cases/user-trust-scores.md)
- [Buyer intent and lead scoring](https://sigwise.ai/docs/use-cases/buyer-intent.md)
- [Chat and content moderation](https://sigwise.ai/docs/use-cases/chat-moderation.md)

## Compared with alternatives

- [SigWise vs. moderation APIs](https://sigwise.ai/docs/compare/moderation-apis.md)
- [SigWise vs. building on an LLM API](https://sigwise.ai/docs/compare/build-it-yourself.md)

## Developers

- [Quickstart](https://sigwise.ai/docs/guide/quickstart.md)
- [API reference](https://sigwise.ai/docs/reference.md), generated from the [OpenAPI 3.1 specification](https://sigwise.ai/docs/openapi.json)
- [SDKs](https://sigwise.ai/docs/sdks.md) for Node.js, Python, Go and PHP
- [Changelog](https://sigwise.ai/docs/changelog.md)

## FAQ

### What is SigWise?

SigWise is an API that reads the events and messages about the objects on your platform (users, listings, orders) and answers questions you define about each of them, such as "is this a scammer?", "how far can we trust them?" or "are they ready to buy?". Each answer is a typed value with probabilities, not free text, so your code can act on it.

### How is it different from a content moderation API?

Moderation APIs classify a single piece of text into a fixed list of harm categories. SigWise answers questions you write yourself, and it judges each new message in the context of everything that object has done before, so it can catch a scam that unfolds over a conversation, score trust over time, or classify buying intent.

### Do I need to train a model?

No. A signal is a plain-language question with a type (yes/no, a labelled score, or one of several categories) and criteria describing the answers. Creating or changing one is a single API call, and it applies from the next analysis.

### How fast are the answers?

By default events are analyzed in the background a few seconds after activity settles, and a burst of events is folded into one analysis. When you need a verdict before something is published, send the events with "wait": true and the answer comes back in the same request.

### How much does it cost?

There is no subscription, seat or minimum. Every new account gets $5 of free credit, then you top up a prepaid balance by card. Each analysis is billed at its actual model cost, typically a fraction of a cent; a short marketplace conversation costs about $0.0001.

### How do I get results into my product?

Read the latest answers from the REST API, search objects by them (for example is_scammer >= 90), receive a signed webhook for every completed analysis, or define rules that send a Slack message, an email or a webhook when a condition becomes true.

### Which languages are supported?

Anything that speaks HTTP and JSON. The API is described by an OpenAPI 3.1 specification, and typed SDKs for Node.js, Python, Go and PHP are generated from it, with request signing, retries and webhook verification built in.

### Is my data kept separate from other customers?

Yes. Every account's objects, events, signals and answers are isolated from every other account's, and API requests are authenticated with short-lived tokens signed by your key's secret, which never leaves your server.

## Get started

Create a free account at https://sigwise.ai/register.