# Ordent

> Ordent is your risk team, as a service. Send events (signup, login, payment, withdrawal); every user gets risk tags from the first request to the last, and every tag comes with a suggested action. Choose how much to delegate: Advise, Recommend, or Autopilot. Self-serve, $99 a month, built for digital businesses that move money. Ordent does not do identity verification or AML; it routes only the users who need it to a verification partner.

Ordent turns entropy into order.

## Start here

- [How it works](https://ordent.io/how-it-works): one event in, one decision out. Detect, decide, act.
- [Docs](https://ordent.io/docs): quickstart and install with your AI agent; reference sections for events, tags, policies, modes, webhooks, widget, labels, sandbox and SDKs.
- [Pricing](https://ordent.io/pricing): Signals $99, Service $399, Risk Team from $2,000. Billed per user event, never per verification. 14-day refund, sandbox before you pay.
- [Start](https://ordent.io/start): get a sandbox key.

## Products

- [Signals](https://ordent.io/products/signals): device intelligence, signup intelligence (email, phone, IP), behavior, actor type (human, bot, AI agent).
- [Graph](https://ordent.io/products/graph): user timeline, clusters, labels, the cross-merchant network.
- [Decisions](https://ordent.io/products/decisions): tags and actions, policies and custom tags, Autopilot, step-up widget, ops panel, KYC routing, AI analyst and weekly report.
- [AI agents](https://ordent.io/agents): build with agents (MCP server, llms.txt, runtime calls) and decide about agents (actor tags, verified agents, agent policies).

## Use cases

- [Fake signups](https://ordent.io/use-cases/fake-signups)
- [Multi-accounting and bonus abuse](https://ordent.io/use-cases/multi-accounting)
- [Account takeover](https://ordent.io/use-cases/account-takeover)
- [Card testing and chargebacks](https://ordent.io/use-cases/card-testing)
- [Crypto exchanges](https://ordent.io/industries/crypto)
- [Fintech](https://ordent.io/industries/fintech)

## Insights

- [Why we give you tags, not a score](https://ordent.io/insights/tags-not-scores): A risk score tells you a number. A tag tells you what happened, how sure we are, and what to do next. Here is why Ordent is built around tags, and what that changes for a team without a fraud analyst. (Field Notes, September 23, 2026)

Full text of every post: https://ordent.io/llms-full.txt. Feeds: https://ordent.io/insights/rss.xml, https://ordent.io/insights/feed.json.

## Integration in one line

```ts
import { Ordent } from "@ordent/sdk";
const ordent = new Ordent({ apiKey: process.env.ORDENT_KEY });
const { decision, tags } = await ordent.events.send({ type: "signup", user: { id, email }, device: req.ordentDevice, ip: req.ip });
if (decision === "step_up") return widget.render({ user: id });
```

Decisions: allow, step_up, review, block, kyc. Modes: advise (you act), recommend (default; we queue edge cases), autopilot (we act, reversible, logged). Tag families: email.*, phone.*, ip.*, device.*, velocity.*, cluster.*, promo.*, payment.*, card.*, actor.*, agent.*, network.*, kyc.*, custom.*.

## Trust

- [The network](https://ordent.io/network): what leaves your account and what doesn't. Keyed hashes, anonymity threshold, give-to-get, cascading deletion.
- [Manifesto](https://ordent.io/manifesto) and [About](https://ordent.io/about).
- Security contact: https://ordent.io/.well-known/security.txt

## Optional

- [Design partners / Risk Team tier](https://ordent.io/design-partners)