Osegi
Startups·2 min read

Naver Cafe AI agent: the higher your rank, the deeper its answers

Naver is testing an AI agent inside three Cafe communities. It answers from members' posts, and what it can read depends on your rank in that cafe.

A smartphone showing a Naver Cafe community page with an AI assistant answering a member's question using summarised posts.
Image: 스타트업 생태계의 모든 것 \'플래텀(Platum)\'

Ask the AI agent in the Tesla Korea cafe how much battery a night of camping mode uses in winter, and it digs up members' own posts and comments on exactly that, summarises them and shows the original posts it used. That is the Naver Cafe AI agent, which Naver started testing in a closed beta on 9 November.

The trial runs in three cafes where members share a lot of information: Nailldong (Japan travel), Tesla Korea (electric vehicles) and Sikment (luxury goods). It is open to members aged 14 or older who have joined those cafes. Some members who have visited the cafe get access first, and Naver will widen the group step by step. Depending on the results, the company plans to bring the agent to cafes on other topics.

Why put an AI agent in Naver Cafe?

According to Naver, the value of cafes lies in first-hand experience: what happened on a trip, what someone learned from using a product for years. That kind of knowledge is scattered across posts and comments and is hard to find with ordinary search. The agent finds posts that match the intent of a question and summarises them. When there is no single right answer, it shows differing opinions side by side. Ask the Nailldong agent whether a ryokan in Yufuin really needs a room with a private open-air bath, and you get the members who recommend one next to those who found the shared bath enough.

Does the AI see everything in a cafe?

No. The answer depends on the asking member's cafe rank. The agent follows each cafe's own rules and rank-based reading permissions. Posts and comments their authors allowed to appear in search or Naver services are used under the same standard as Naver Search. Posts that were not made public are used only within what the asking member can already read at their current rank.

So if you become more active and move up a rank, the posts the agent can draw on grow too. In effect, the AI doesn't replace the cafe's rank culture. It gives people one more reason to take part.

Kim Eun-jung, Naver Cafe leader, said the cafe's value lies in deep knowledge on specific topics, hands-on know-how and the relationships built along the way, and that Naver will expand the feature while respecting each community's culture and way of running things.

Apps worth a look

SUMO: the number puzzle

Tap the right tiles, add them up, hit the target. Easy to learn, hard to master. Free, for all ages.

Tags:navernaver cafeai agentclosed betakoreacommunity

More from this day

See all →

More in: Startups

See all →