Korea's search market does not run through Google; it runs through Naver, and Naver now answers many queries with an AI Briefing block before any organic result appears. If your brand sells in Korea, that answer, and the sources it cites, is the visibility that matters. I'll cover the manual way to monitor it, where that falls apart, and the shortcut: the Naver AI Overview API on Apify, which returns the briefing, its citations, and related media as structured JSON, one row per query.
Disclosure: the Apify links in this post are affiliate links. If you run the Actor, I may earn a referral commission at no extra cost to you.
Does Naver have an API for the AI Briefing?
Naver Developers ships open APIs for blog, news, and shopping search, and none of them return the AI Briefing block. The briefing renders inside the search page, shifts as Naver retunes it, and has no documented endpoint. So monitoring it means scraping it, ideally through something that behaves like an API: a query goes in, a structured answer comes out.
What the Naver AI Overview API returns
The Naver AI Overview API returns one JSON row per query: whether an AI Briefing was shown, the full AI-generated answer, parsed text blocks with reference indexes, the cited sources, related media, and follow-up questions.
| Data point | Example | Notes |
|---|---|---|
| Overview shown | true | some queries get no briefing, and the row says so |
| AI answer | full text | original formatting preserved |
| Parsed blocks | text blocks with reference indexes | maps claims to their citations |
| Cited sources | titles, links, snippets, source names | the signal AEO work runs on |
| Related media | videos and images with metadata | when Naver attaches them |
| Follow-up questions | related questions | what Naver suggests asking next |
Who this is for
Korean AEO and brand teams checking whether they appear in AI answers, agencies auditing competitor visibility across keyword sets, and builders feeding the current Korean answer into an agent or a dashboard.
The manual way, and where it breaks
The manual loop is searching each keyword on Naver, screenshotting the briefing, and pasting citations into a sheet, then repeating next week. It fails on volume and on consistency: the block appears for some queries and not others, the answer text changes without notice, and the citations, the part you actually care about, are the fiddliest bits to copy. A DIY script faces a JavaScript-rendered Korean page whose layout Naver adjusts freely.
The faster way: run the Naver AI Overview API
Apify Console
- Open the Naver AI Overview API and click Try for free.
- Enter a query such as
당뇨병 증상(diabetes symptoms), or paste a whole list underqueries. - Run it and export the rows as JSON or CSV.
REST
curl -X POST "https://api.apify.com/v2/acts/johnvc~naver-ai-overview-api/runs?token=YOUR_APIFY_TOKEN" \
-H "Content-Type: application/json" \
-d '{ "query": "당뇨병 증상" }'
The run lifecycle is documented in the Apify API docs.
Monitor Korean AI answers in Python
from apify_client import ApifyClient
client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("johnvc/naver-ai-overview-api").call(
run_input={
"queries": ["선크림 추천", "쿠션 파운데이션 추천"],
}
)
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
print(row)
Read each row's shown flag first, then the cited source list; brand monitoring is mostly the question of whether your domain is in that list.
Check if your brand appears in the AI Briefing
Check if your brand appears in Naver AI briefing runs your brand and category keywords and lets you scan answers and citations for mentions.
Batch-monitor whole keyword sets
Check Naver AI briefing for many keywords runs a list in one go, and Monitor Naver AI briefing for K-beauty keywords shows the same pattern on a live vertical.
See which sources Naver cites
See which sources Naver AI briefing cites flips the question from "do we appear" to "who does Naver trust for this category", which tells you where to pitch or publish.
Export to CSV or wire up GEO tracking
Export Naver AI briefing to CSV lands the rows in a spreadsheet, and Naver AI briefing API for GEO tracking in Korea feeds generative-engine-optimization reporting.
Use it from Claude via MCP
Over the Model Context Protocol, the Actor becomes a tool Claude, Claude Code, or Cursor can call, so an agent can fetch the current Korean answer for a keyword mid-conversation. The setup is in Check Naver AI briefing from Claude via MCP, and you can get Claude itself at claude.ai.
FAQ about scraping Naver's AI Briefing
What does the Naver AI Briefing scraper cost per query?
About 1.8 cents per query resolved on the free tier, so a 100-keyword monitoring run stays under two dollars. Apify's free platform credit typically covers the first experiments.
Does every query give the scraper a briefing to collect?
No. Naver shows AI Briefings mostly for informational, question-style searches; when none appears, the row records that. I treat those misses as data too, since a keyword gaining or losing its briefing is exactly what you want to catch.
What is AEO, and why point a scraper at Naver for it?
Answer engine optimization means making sure AI-generated answers mention and cite you. Naver is South Korea's largest search engine, and its AI answers increasingly shape what Korean users see first, so Korean AEO starts with knowing what those answers actually say.
Can Claude run this scraper through MCP?
Yes. Connected via the Apify MCP server, it appears as a callable tool in Claude, Claude Code, and Cursor, taking the same query inputs.
Can I schedule the scraper to catch answer changes?
Yes, and that is the intended shape: repeated runs on the same keyword set surface changes in answers and citation patterns over time. Save a task and attach a schedule from the Naver AI Overview API.
More from Truffle Pig Data
Related monitoring Actors: the Naver Search API for organic Naver results, the Google AI Overview API for the same job on Google, and the Bing Copilot API for Microsoft's answers.
Wrapping up
If Korea matters to your brand, the AI Briefing is the new first impression, and it is checkable by machine. Load your keywords into the Naver AI Overview API and see what Naver is telling your customers.
Top comments (0)