# Recommend and similar

Given a product, return related products. Recommendations are computed live from the product graph, so they stay fresh.

```bash
curl "https://dev.ziffi.xyz/v1/recommend?product_id=ba52542f-d40c-48e0-a579-5e0bfc80b3dd&intent=related&limit=12"
```

## Intents

| `intent` | Returns |
|---|---|
| `related` (default) | Nearest overall, good for "you may also like". |
| `cross_brand` | Nearest products from other brands, good for comparison. |

```bash
curl "https://dev.ziffi.xyz/v1/recommend?product_id=…&intent=cross_brand"
```

## The response

The `data` is an array of the same product cards you get from search, so you can render them with the same component.

```json
{
  "data": [ { "id": "…", "title": "…", "brand": "…", "price_min_minor": 34900, "currency": "INR" } ],
  "intent": "related",
  "count": 12,
  "tier": "graph"
}
```

## Next steps

- [Search products](#/search): the same card shape.
- [Build an AI agent](#/agents): recommend over MCP.