> ## Documentation Index
> Fetch the complete documentation index at: https://docs.zhentan.me/llms.txt
> Use this file to discover all available pages before exploring further.

# Introduction

> Your personalized onchain detective and assistant that learns and guards your onchain behavior.

Zhentan is a personalized wallet assistant that acts as an AI co-signer on a Safe multisig **where you always hold the majority of keys**. In the full 2-of-3 model — your embedded key, your backup key, and the agent — it learns your behavioral patterns and screens every transaction before it executes: blocking threats, flagging anomalies for review, and auto-approving routine activity. And because your two keys meet the signing threshold, you can always execute from the official Safe app without the agent.

<CardGroup cols="2">
  <Card title="Zhentan App" icon="shield-check" iconType="duotone" href="/app/overview">
    Send, receive, and connect to DApps with AI-powered transaction screening on every action.
  </Card>

  <Card title="Technology" icon="microchip" iconType="duotone" href="/technology/overview">
    Deep dive into the architecture — the 2-of-3 signing model, Safe Transaction Service, risk scoring, and the NanoBot/Hermes agent.
  </Card>

  <Card title="Quickstart" icon="rocket-launch" iconType="duotone" href="/quickstart">
    Run Zhentan locally with full AI screening in minutes.
  </Card>

  <Card title="API Reference" icon="code" iconType="duotone" href="/api/introduction">
    Integrate with the Zhentan server API for queue management, execution, and portfolio data.
  </Card>
</CardGroup>

## Why Zhentan?

DeFi wallets offer no protection between "sign" and "execute." Once you click approve, the transaction goes through — even if it's a scam, a fat-finger, or a social engineering attack.

* **\$2.2B+** lost to crypto hacks and scams in 2023 alone (Chainalysis)
* Phishing, address poisoning, and honeypot tokens are the top attack vectors
* Hardware wallets and browser extensions only confirm what you already signed — they don't screen it

## How It Works

Zhentan adds an AI co-signer to your Safe wallet. Every transaction is a standard SafeTx — visible in [app.safe.global](https://app.safe.global) from the moment you sign — scored 0–100 against your learned behavioral patterns before anything executes. The agent relays execution and pays gas; it never signs a transaction it didn't screen, and it can never move funds alone.

| Score | Verdict     | Action                                       |
| ----- | ----------- | -------------------------------------------- |
| \< 40 | **APPROVE** | Auto-signed and executed immediately         |
| 40–70 | **REVIEW**  | Deep analysis + Telegram interactive buttons |
| > 70  | **BLOCK**   | Rejected with alert                          |

## Live Demo

<Note>
  [zhentan.me](https://zhentan.me) runs with a public NanoBot/Hermes agent — no personalized screening. For full AI screening (APPROVE / REVIEW / BLOCK), [run locally](/quickstart) with the server and agent.
</Note>