Search

Xiaohongshu (RED) Content Research with AI Agents

Use one AI agent to run Xiaohongshu (RED) content research: expand topics from trending and autocomplete terms, break down top-performing notes, and mine comments for real audience needs — producing a data-backed topic list with angles.

Xiaohongshu (RED) Content Research with AI Agents

Tool names in this guide are for reference only and may have changed or been renamed. Please use the latest tools in the tool marketplace when implementing.

Key takeaways

  • Go from trending topics and keyword hints to comment insights for topic selection
  • Base topics on data instead of hunches
  • Hand insights straight into image/copy generation

Tools used

Why Xiaohongshu topics always feel like guesswork

On Xiaohongshu (RED), the scariest part of content is "picking topics on a hunch." You write with no data behind it, then find out after posting that nobody cares. You want to ride a trend, but you don't know what the platform is actually searching for right now. And even when you find a few viral notes, you just imitate them on feel, without being able to say why they worked.

The root problem: the information you need for topic research lives in several separate places — trending topics in one spot, search autocomplete in another, the angles and structure of reference notes you have to open one by one, and the real audience needs buried in the comments. Stitching all of that together by hand is slow and fragmented, and you usually end up back on "gut feel."

A better way is to let one AI agent run the whole "trending → autocomplete → notes → comments" chain in a single pass, reverse-engineering topics from real data instead of starting from a blank document.

The workflow in four steps

  1. Give a niche — tell the agent one niche or topic keyword.
  2. Expand topics — pull candidate directions from trending topics and search autocomplete.
  3. Break down top notes — search reference notes and dissect their angles and structure.
  4. Mine needs — extract real pain points and frequent questions from the comments.

The agent then rolls it up into a topic list with concrete angles.

How to do each step

Step 1: Give the agent a niche keyword

Just say: Use ae to run a round of Xiaohongshu topic research around the 'camping gear' niche. The agent uses that as the seed for the data retrieval that follows.

Step 2: Expand topics with trending + autocomplete

The agent first uses a Xiaohongshu trending tool (such as ae_xiaohongshu_hot_search) to see what the platform is buzzing about, then a keyword-recommendation tool (such as ae_xiaohongshu_keyword_recommend) to get the search autocomplete terms for your niche. Trending tells you which way the wind is blowing; autocomplete tells you how users actually phrase their searches. Together they turn a broad niche into dozens of concrete topic directions.

Step 3: Dissect reference notes

With candidate directions in hand, the agent uses a note-search tool (such as ae_xiaohongshu_note_search) to surface high-performing reference notes for each direction, and analyzes their title hooks, content structure, and angles. The point here is to understand how the angle and structure are built — not to copy content.

Step 4: Mine comments for real needs

The comment sections under viral notes are a gold mine. The agent uses a note-comment tool (such as ae_xiaohongshu_note_comment_list) to pull comments from reference notes and cluster them into "what people are still unsure about" and "what they keep asking." Those real questions become the hooks and FAQs for your next note.

Aspect Topics on gut feel Topics from trending + comment data
Topic source Personal inspiration Platform trending + autocomplete
Benchmarking Vague imitation Dissecting structure and angle
Need insight Guessing what users want Real questions in comments
Hit rate Luck-based Data-backed

Example topic output (neutral placeholder data):

Candidate topic Data source Angle
Beginner camping checklist pitfalls High-frequency autocomplete Comments keep asking "should I buy XX"
Low-budget camping outfits Trending-related Reference note uses "a full set for $XXX"

How to get started

  1. Sign up for AgentEarth — new users get 1,000 free credits.
  2. In agents like WorkBuddy or Trae, search for "AgentEarth" and connect in-app; or paste a one-line MCP config into Claude Code / Cursor / VSCode.
  3. Give instructions in natural language, e.g. Use ae to run a round of Xiaohongshu topic research for a niche.

FAQ

Can I get trending topics and autocomplete terms? Yes. A trending tool (such as ae_xiaohongshu_hot_search) and a keyword-recommendation tool (such as ae_xiaohongshu_keyword_recommend) cover platform trending and search autocomplete respectively, and the agent combines both to expand topics.

Can comments be analyzed in bulk? Yes. The agent pulls comments from reference notes with a note-comment tool and clusters them into shared needs and frequent questions, so you don't read them one by one.

Can I go straight from topic to a finished post? You can chain it. Once the topic is set, you can have the agent call image-generation tools to produce visuals, closing the "topic → finished post" loop.


Sign up for 1,000 credits · Browse 2,400+ tools · Integration guide