Oxylabs AI Search

Query the major AI answer engines and get back the AI-generated answer text plus its cited source URLs for GEO/AEO visibility monitoring.

POST https://api.aisa.one/apis/v1/oxylabs/ai-search

cURL

google_search

curl --request POST \
  --url https://api.aisa.one/apis/v1/oxylabs/ai-search \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "source": "google_search",
  "query": "best noise cancelling headphones 2026",
  "parse": true,
  "render": "html",
  "geo_location": "United States"
}
'

200

{
  "results": [
    {
      "content": {
        "results": {
          "ai_overviews": [
            {
              "answer_text": [
                {
                  "fragments": [
                    {
                      "references": [
                        "<unknown>"
                      ]
                    }
                  ]
                }
              ],
              "references": [
                {
                  "source": "<string>",
                  "url": "<string>"
                }
              ]
            }
          ]
        }
      }
    }
  ]
}

Run a real query against a major AI answer engine and get back the AI-generated answer_text together with the cited source URLs — built for GEO/AEO, so you can monitor how a brand or product is surfaced and cited across AI answers. Pick the engine with source (chatgpt, gemini, perplexity, google_search for Google AI Overviews, or google_ai_mode) and send the parameters that source expects: prompt for chatgpt/gemini/perplexity, query with render: "html" for the Google-type sources, plus parse: true and a country-level geo_location such as "United States".

The request body is passed through unchanged to the upstream engine, and the parsed shape inside each results[] entry varies by source — google_search returns ai_overviews[] with answer_text and references[]{source, url} (the shape documented here), while chatgpt/gemini return response_text + citations[], perplexity returns top_sources[]/sources_results[], and google_ai_mode returns content.citations[]{text, urls[]}. Billed at a flat $0.001 per successful result; 400/429/5xx/6xx and upstream 4xx responses are not billed. AI sources (chatgpt/gemini/perplexity) take ~40–60s and Google-type sources ~4–8s, so set your client timeout to at least 90s.

Authorizations

Authorization · string · header · required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Body

application/json

Passthrough body. source selects the AI answer engine; the remaining fields are the parameters that source expects (see per-source notes on each property). Additional upstream parameters are passed through as-is.

source · enum<string> · required

The AI answer engine to query. google_search returns Google AI Overviews. Each source expects a specific subset of the parameters below.

Available options: chatgpt, gemini, perplexity, google_search, google_ai_mode

Example:

"google_search"

prompt · string

The natural-language prompt. Used by chatgpt (max 4000 chars), gemini (max 8000 chars), and perplexity. Use query instead for the Google-type sources.

Example:

"best noise cancelling headphones 2026"

query · string

The search query. Used by google_search and google_ai_mode. Use prompt instead for chatgpt/gemini/perplexity.

Example:

"best noise cancelling headphones 2026"

search · boolean

For chatgpt, set to true to have ChatGPT browse the web before answering.

Example:

true

render · enum<string>

For google_search and google_ai_mode, set to "html" to render the page before parsing.

Available options: html

Example:

"html"

parse · boolean

Return structured, parsed results instead of raw output. Recommended for every source.

Example:

true

geo_location · string

Country-level geo-location for the query, e.g. "United States".

Example:

"United States"

Response

200 - application/json

Query completed successfully. results[] holds one result per query. The parsed shape inside content varies by source (see property descriptions); the example below shows the google_search (Google AI Overviews) shape.

results · object[]

One entry per query. A Realtime single query returns exactly one result, which is the billed unit.

Show child attributes