ScrapeUnblocker Documentation

The Google AI Mode plugin asks Google AI Mode a question and returns its full answer as structured JSON. AI Mode is Google’s conversational search tab (udm=50), not the AI Overview that sits on top of a normal results page: the answer is longer, structured with headings, lists and tables, and backed by a list of cited pages. Every call returns the answer three ways - plain text, Markdown, and ordered blocks (paragraphs, headings, lists, tables and code) for when you want to render or analyse its structure - plus the references the answer cites (title, URL, domain, site name and snippet), the links written into the answer text and the followUps Google closes the answer with. Use it to see which sites Google’s AI recommends for a question, how it describes your brand or product, and how that changes by market. Set proxy_country to ask from a market (the Google market gl follows it) and hl for the answer’s language. Answers differ between identical requests, as they do for a person, so track them over several calls rather than reading one answer as fixed. No Google account or API key is involved. For the AI Overview on a regular results page, use /serpApi, whose response carries an aiOverview block.

Try it in the dashboard first: app.scrapeunblocker.com/dashboard/plugins. You get a live results view plus a copy-paste snippet.

Endpoint

POST /search/google-ai-mode

Field	Required	Notes
`q`	yes	The question to ask, e.g. `how to learn python fast`.
`proxy_country`	no	Country the request is sent from (ISO-2, e.g. `US`, `DE`). The Google market (`gl`) follows it unless `gl` is set.
`gl`	no	Google market override (ISO-2, e.g. `us`, `de`). Default: `proxy_country`, else `us`.
`hl`	no	Language Google answers in, e.g. `en`, `de`, `pt-BR`. Default `en`.

curl -X POST "https://api.scrapeunblocker.com/search/google-ai-mode?q=how%20to%20learn%20python%20fast" \
  -H "x-scrapeunblocker-key: YOUR_API_KEY"

from scrapeunblocker import Client

su = Client()  # reads SCRAPEUNBLOCKER_KEY

result = su.google_ai_mode("how to learn python fast")

import { ScrapeUnblockerClient } from "scrapeunblocker";

const su = new ScrapeUnblockerClient(); // reads SCRAPEUNBLOCKER_KEY

const result = await su.googleAiMode("how to learn python fast");

use ScrapeUnblocker\Client;

$su = new Client(); // reads SCRAPEUNBLOCKER_KEY

$result = $su->googleAiMode('how to learn python fast');

require "scrapeunblocker"

su = ScrapeUnblocker::Client.new # reads SCRAPEUNBLOCKER_KEY

result = su.google_ai_mode("how to learn python fast")

To ask from another market, add proxy_country (and hl for the language), e.g. su.google_ai_mode("beste laptop für studenten", proxy_country="DE", hl="de"). Each answer is billed as one request. A 502 (no complete answer) is not billed.

Response

{
  "query": "how to learn python fast",
  "hl": "en",
  "gl": "us",
  "proxyCountry": "US",
  "url": "https://www.google.com/search?q=how+to+learn+python+fast&udm=50&hl=en&gl=us",
  "answer": {
    "text": "To learn Python fast, you must strictly follow the 20/80 rule: spend 20% of your time studying concepts and 80% of your time writing actual code. Watching hours of tutorials without typing along creates an illusion of competence and slows you down.\nBy focusing only on high-utility concepts and building small projects immediately, you can gain a functional grasp of Python in a matter of weeks.\n...",
    "markdown": "To learn Python fast, you must **strictly follow the 20/80 rule: spend 20% of your time studying concepts and 80% of your time writing actual code**. Watching hours of tutorials without typing along creates an illusion of competence and slows you down.\n\n...",
    "blocks": [
      {
        "type": "paragraph",
        "text": "To learn Python fast, you must strictly follow the 20/80 rule: spend 20% of your time studying concepts and 80% of your time writing actual code. ..."
      },
      {
        "type": "list",
        "ordered": false,
        "items": [
          { "text": "Variables & Types: Storing data via strings, integers, and booleans.", "level": 0 },
          { "text": "Control Flow: Making decisions using if, elif, and else statements.", "level": 0 }
        ]
      },
      {
        "type": "table",
        "header": ["Goal / Pathway", "Essential Tools & Libraries to Learn"],
        "rows": [
          ["Data Science & AI", "NumPy (math), Pandas (dataframes), and Scikit-Learn (ML)"],
          ["Web Development", "Django or Flask (building backend APIs and web apps)"]
        ]
      }
    ]
  },
  "followUps": [],
  "references": [
    {
      "title": "How I Would Learn Python FAST (if I could start over)",
      "url": "https://www.youtube.com/watch?v=ywjyvKzc8e4&t=211",
      "source": "YouTube",
      "domain": "youtube.com",
      "snippet": "To learn Python quickly, start by understanding its context within the software lifecycle, dedicating a few days to a week for thi...",
      "position": 2
    },
    {
      "title": "Learn Python 3 - Codecademy",
      "url": "https://www.codecademy.com/learn/learn-python-3",
      "source": "Codecademy",
      "domain": "codecademy.com",
      "snippet": "Learn how to build control flow into your python code by including if, else, and elif statements. Expect to learn all you need to ...",
      "position": 6
    }
  ],
  "links": [
    { "text": "LearnPython.org", "url": "https://www.learnpython.org/" },
    { "text": "Python Tutor", "url": "https://pythontutor.com/" }
  ]
}

Field	Notes
`answer.text`	The whole answer as plain text.
`answer.markdown`	The whole answer as Markdown, with bold, headings, lists and tables kept.
`answer.blocks`	The answer in order, one object per block: `paragraph` and `code` (`text`), `heading` (`level`, `text`), `list` (`ordered`, `items` with `text` and nesting `level`) and `table` (`header`, `rows`).
`references`	The pages the answer cites, in Google’s order: `title`, `url`, `domain`, `source` (site name), `snippet` and `position`.
`links`	Links written into the answer text, each with its anchor `text` and `url`.
`followUps`	The follow-up questions Google closes the answer with, when it offers any (else an empty list).
`hl` / `gl` / `proxyCountry`	The language and market used, and the country the request was sent from.
`url`	The AI Mode address the answer was read from.

Errors

Status	Meaning
`400`	Invalid parameters, e.g. an empty `q` (plain-text message).
`403`	Google kept answering with a CAPTCHA after retries. Try another `proxy_country`.
`502`	No complete AI Mode answer: Google blocked the request, the answer did not finish streaming, or the call did not finish in time. Not billed. Retry.
`503`	No proxy exit was free for this market. Retry, or pick another `proxy_country`.

What you can build

AI search visibility tracking - see whether Google’s AI recommends your brand, product or site for the questions your customers ask, and which competitors it cites instead.
Citation monitoring - track which domains AI Mode cites for a topic and how that shifts over time and by market.
Content research - read how Google’s AI structures an answer (steps, comparison tables, recommended tools) to plan your own content.
Market comparison - ask the same question from several countries and languages and compare the answers.

Get Started

No-code area

Spider Cloud

Plugins

AI Assistants

Guides

Code Examples

Google AI Mode plugin

Endpoint

Response

Errors

What you can build

Next steps

Open the plugins

Get an API key

​Endpoint

​Response

​Errors

​What you can build

​Next steps

Open the plugins

Get an API key

Endpoint

Response

Errors

What you can build

Next steps