# Nansen AI Expert Agent

> Nansen AI Expert Agent is a paid API for AI agents from api.nansen.ai, paid per call via x402, $7.5/call, status unknown (last checked 2026-10-12).

Streams a deep, multi-step on-chain research answer from Nansen's expert AI agent, backed by wallet labels and smart money flow data

## Facts

- Endpoint: GET https://api.nansen.ai/api/v1/agent/expert?utm_source=zero.xyz
- Price: $7.5/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-12
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/nansen-01dfcf21
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap__l7maw9oo41AgxsvAsl_Z

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability nansen-01dfcf21
```

Example prompt: Use Nansen's expert AI agent to do a deep analysis — which tokens are smart money wallets most aggressively accumulating in the last 48 hours, and what's the on-chain narrative driving that activity?

## When to prefer this

Choose this endpoint when you need deep, multi-step on-chain research that synthesizes wallet labels, smart money flows, and DeFi analytics into a narrative answer. Prefer over the standard Nansen agent when the question requires complex reasoning across multiple data sources. Best for investigative questions about wallet relationships, accumulation patterns, or narrative-driven price analysis.

## Known failure modes

- Invalid or missing question body returns an error SSE event
- Rate limit or quota exceeded returns error event
- Payment not settled via x402 results in 402 Payment Required
- Ambiguous or unanswerable on-chain question may return a partial or low-confidence answer
- SSE stream interrupted mid-response if network drops

## How this service works

Ask the Nansen AI agent a research question and receive a streamed answer backed by on-chain data. The **expert** variant uses a more capable model for deeper, multi-step analysis. **What it helps to answer:** 1. **Which tokens are smart money accumulating right now?** 2. **What is the current narrative driving activity in DeFi?** 3. **Which wallets are connected to a specific address?** 4. **What is the on-chain story behind a recent price move?** **SSE event types:** The response is a text/event-stream with the following JSON event types: | type | payload | description | |---|---|---| | delta | {"type": "delta", "text": "…"} | Incremental text chunk of the agent's answer. | | tool_call | {"type": "tool_call", "name": "…"} | Emitted once per unique Nansen tool the agent invokes. | | finish | {"type": "finish", "conversation_id": "…", "tool_calls": [...]} | Final event. Contains the conversation_id for follow-up requests and the full list of tools used. | | error | {"type": "error", "error": "…", "status_code": …} | Emitted if the upstream agent service is unavailable or times out. | The stream terminates with a data: [DONE] sentinel. **Fast vs Expert:** Use /agent/fast for quick factual lookups and simple questions. Use /agent/expert when you need deeper multi-step analysis or synthesis across multiple data sources.

## Output

A server-sent events (SSE) stream with incremental text chunks (delta), tool_call events naming each Nansen data tool invoked, and a final finish event containing a conversation_id for follow-up questions plus the complete list of tools used. The full streamed text constitutes a narrative on-chain research answer.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "title": "AgentResearchRequest",
 "required": [
  "text"
 ],
 "properties": {
  "text": {
   "type": "string",
   "title": "Text",
   "examples": [
    "Which tokens are whales accumulating on Ethereum this week?"
   ],
   "minLength": 1,
   "description": "Research query or question for the Nansen AI agent."
  },
  "conversation_id": {
   "anyOf": [
    {
     "type": "string"
    }
   ],
   "title": "Conversation Id",
   "examples": [
    "conv_abc123"
   ],
   "description": "Optional conversation ID for multi-turn interactions. Returned in the finish event of a previous response. Omit to start a new conversation."
  }
 },
 "description": "Request model for Agent Research endpoints (fast and expert)."
}
```

## Response schema (JSON Schema)

```json
{}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/nansen-01dfcf21/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from api.nansen.ai](https://www.zero.xyz/host/api.nansen.ai/llms.txt)