---
title: "Visual Search"
method: POST
path: "/api/v1/intelligence/visual-search"
tags: ["Intelligence"]
---

# Visual Search

`POST /api/v1/intelligence/visual-search`

## Request

- Base URL: `https://api.casaintelligence.com.au` (the document declares no server; this is the origin it was published from)
- URL: `https://api.casaintelligence.com.au/api/v1/intelligence/visual-search`
- Auth: HTTP bearer

Multimodal visual document search using pgvector cosine similarity.

Searches only image-embedded chunks (chunk_index < 0) to find visually
similar document pages. Text queries work because text and image embeddings
share the same 768D vector space via Gemini Embedding 2.

## Request body

- VisualSearchRequest: Request body for multimodal visual document search.
  - `query` string, required: Natural language search query
  - `lga` string, nullable: Filter by LGA
  - `document_type` string, nullable: Filter by document type
  - `limit` integer: Max results to return

## Response `200`

Successful Response

- VisualSearchResponse: Response for visual document search.
  - `results` VisualSearchResult[]
    - `document_id` string, required
    - `page_number` integer, required
    - `document_type` string, nullable
    - `similarity_score` number, nullable
    - `application_number` string, nullable
    - `address` string, nullable
    - `lga` string, nullable
    - `filename` string, nullable
    - `da_id` string, nullable
  - `total_count` integer
  - `query` string

## Other responses

- `422`: Validation Error

---

[API](https://skmtc.dev/casaintelligence/apis/casa-intelligence-engine.md) · [All operations](https://skmtc.dev/casaintelligence/apis/casa-intelligence-engine/llms.txt) · [OpenAPI document](https://skmtc.dev/casaintelligence/apis/casa-intelligence-engine/revisions/71d711039de0?raw)