What Platforms Are the Data Sources for AI Research?

Question Type

User Manual Question

Quick Answer

Social Media Scanning Supported Platforms:

🇨🇳 Domestic: Xiaohongshu, Douyin
🌍 Overseas: Twitter, Instagram, TikTok

AI Research (Interview/Discussion):

Uses atypica's 300K+ AI persona library, doesn't depend on social media data
AI personas based on: Real interview transcripts + Real user profiles + Social media scanning generation

Data Characteristics:

✅ Only scans public content, no privacy concerns
✅ Real-time latest, not historical data
✅ Can specify platforms and regions

Social Media Scanning Supported Platforms

Domestic Platforms (2)

1. Xiaohongshu

Characteristics:

User Demographics: Primarily young women (70%+ female, 18-35 years old)
Content Types: Consumer product reviews, lifestyle sharing, beauty & skincare
Discussion Style: Real experience sharing, image and text

Suitable for Scanning:

✅ Consumer goods (beauty, skincare, food, apparel)
✅ Lifestyle (fitness, travel, home)
✅ Young women's consumer psychology
✅ Product reviews and genuine feedback

Example:

Scanning Target: "25-35 year old women's attitudes toward 0-sugar snacks"Recommended Platform: Xiaohongshu (First choice)Reason: Xiaohongshu users highly overlap, discussions about real experiences

2. Douyin

Characteristics:

User Demographics: Wide range (18-50 years old), good lower-tier market coverage
Content Types: Short videos, live streaming e-commerce, life records
Discussion Style: Fragmented, highly entertaining

Suitable for Scanning:

✅ FMCG and daily necessities
✅ Lower-tier market users
✅ Live streaming e-commerce related
✅ Short video content trends

Example:

Scanning Target: "Lower-tier market users' attitudes toward Pinduoduo"Recommended Platform: DouyinReason: Lower-tier market users active

Overseas Platforms (3)

1. Twitter

Characteristics:

User Demographics: Global users, tech and business circles
Content Types: Real-time discussions, opinion sharing, news dissemination
Discussion Style: Fast, concise, clear opinions

Suitable for Scanning:

✅ Overseas market user attitudes
✅ Tech product discussions
✅ Business trends
✅ Public opinion hotspots

Example:

Scanning Target: "Overseas users' attitudes toward AI tools"Recommended Platform: TwitterReason: Tech circle gathering place, active discussions

2. Instagram

Characteristics:

User Demographics: Global users, primarily young people
Content Types: Images, short videos, Stories
Discussion Style: Visual, lifestyle showcase

Suitable for Scanning:

✅ Fashion and beauty
✅ Lifestyle brands
✅ Visual products
✅ Overseas young people consumption

Example:

Scanning Target: "Overseas young people's attitudes toward sustainable fashion"Recommended Platform: InstagramReason: Fashion enthusiasts gathering place

3. TikTok

Characteristics:

User Demographics: Global young users (18-35 years old)
Content Types: Short videos, creative content, life sharing
Discussion Style: Highly entertaining, highly interactive

Suitable for Scanning:

✅ Young people consumption trends
✅ Product seeding and reviews
✅ Creative marketing content
✅ Global market insights

Example:

Scanning Target: "Overseas young people's attitudes toward certain beauty brand"Recommended Platform: TikTokReason: Young users active, product reviews genuine

Platform Selection Recommendations

Choose by Demographics

Young Women (18-35 years old):

First Choice: Xiaohongshu
Alternative: Douyin

Gen Z (18-25 years old):

First Choice: Douyin
Alternative: Xiaohongshu

Lower-Tier Market:

First Choice: Douyin

Overseas Market:

First Choice: Twitter, Instagram
Alternative: TikTok

Choose by Product Type

Consumer Goods (Beauty, Food, Apparel):

Xiaohongshu, Douyin, Instagram

Lifestyle (Fitness, Travel, Home):

Xiaohongshu, Instagram

FMCG and Daily Necessities:

Douyin, TikTok

Multi-Platform Cross-Validation

Recommended Combinations:

Combination 1: Young women consumer goods

Xiaohongshu + Douyin- Xiaohongshu: Deep experience sharing- Douyin: Quick seeding and e-commerce

Combination 2: Overseas market

Twitter + Instagram + TikTok- Twitter: Real-time discussions- Instagram: Visual content- TikTok: Short video reviews

Value:

Cross-validate findings
Avoid platform bias
More comprehensive user insights

AI Research Data Sources (Non-Social Media)

Core Source: 300K+ AI Persona Library

AI personas are not real-time scraped from social media, but pre-built:

Persona Sources:

Real Interview Transcripts:
- Real users' interview content
- Genuine attitudes, opinions, decision processes
- Converted into structured AI personas
Real User Profiles:
- Demographic information
- Psychological characteristics
- Behavioral patterns
Social Media Scanning Generation:
- Scout → Generate custom personas
- Built based on real user content

AI Persona Quality:

Custom Personas: Consistency reaches 85 points (exceeds real human baseline 81 points, built from social media scanning/interview transcripts)
Public Persona Library: Consistency reaches 80 points (close to real human baseline 81 points, built from general profiles)
Temporary Generation: For quick validation

Interview/Discussion Doesn't Depend on Social Media

Difference:

Social Media Scanning:- Real-time scan social media- Extract real user content- Generate demographic profile and AI personasAI Interview/Discussion:- Use existing AI persona library- AI personas simulate real user feedback- Don't depend on social media real-time data

Example:

Scenario: Test sparkling coffee new productStep 1: Social Media Scanning (Uses social media)- Scan Xiaohongshu/Douyin discussing sparkling coffee- Generate demographic profile + 3 custom AI personasStep 2: AI Interview (Doesn't use social media)- Use 3 AI personas for deep conversations- AI personas answer based on own persona background- Won't check social media real-time data

Data Privacy and Security

Only Scan Public Content

Social Media Scanning:

✅ Only scan publicly available social media content
❌ Don't access any private information
❌ Don't access account private messages or private content

Public Content Definition:

Xiaohongshu public posts
Douyin public content
Other social platform public content
Other platforms' publicly posted content

Don't Store Original Personal Information

Processing Flow:

Step 1: Scan public content- Extract content: User-posted text, images (no personal identity info)Step 2: AI analysis- Extract opinions, attitudes, behavioral patterns- Don't record specific user IDs or accountsStep 3: Generate insights- Output: Demographic profile, behavioral patterns- Don't output: Specific user identity information

Example:

❌ Won't Record:- "@Zhang San says: I like coffee"- "Xiaohongshu user ID: 123456"✅ Will Extract:- "Young women users discussing coffee frequently mention 'appearance' and 'photo-worthy'"- "Price acceptance: ¥25-28 optimal"

Complies with Platform Policies

Compliance:

✅ Follow each platform's crawler policies
✅ Only access public APIs or public pages
✅ Don't violate any platform's user agreements

Real-Time and Data Freshness

Real-Time Scanning

Social Media Scanning:

Scan latest social media content
Not historical database
Reflects current user attitudes

Example:

January 2026 Scanning:- Discovered "emotional value" becoming new selling point for healthy snacks- This is latest trend, industry reports may not cover yetvs Traditional Industry Reports:- Usually lag 3-6 months- May miss latest trends

Regular Update Recommendations

Market is Changing:

Recommend re-scan every 3-6 months
Discover trend changes
Adjust strategy promptly

Example:

October 2025 Scanning:- Healthy snack users focus on "low sugar low fat"January 2026 Re-Scan:- Users start focusing on "emotional value" and "healing feeling"- Trend shift: Functional health → Emotional healthRecommendation:- Adjust product positioning and marketing strategy

FAQ

Q1: Will All Platforms Be Scanned Simultaneously?

No, select based on task:

Default: AI recommends 2-3 platforms based on demographic characteristics Manual Specification: Can specify specific platforms

Example:

Scanning Target: "25-35 year old women's attitudes toward healthy snacks"AI Recommends:- Xiaohongshu (Primary): Young women gathering place- Douyin (Secondary): Quick seeding- Douyin (Optional): FMCG seedingReason:- Xiaohongshu users best match- Douyin broader coverage, good seeding effect- Douyin can supplement lower-tier market perspective

Q2: Can Scan Only Specific Regional Users?

Yes:

Setting Method:

Scanning Target: "Users in Chengdu region discussing coffee consumption"Scout will:- Focus on users in Chengdu region- Analyze regional characteristics- Output Chengdu user profile

Supported Regional Filtering:

First-tier cities (Beijing, Shanghai, Guangzhou, Shenzhen)
New first-tier cities (Chengdu, Hangzhou, Wuhan, etc.)
Provinces (Sichuan, Zhejiang, etc.)
Overseas regions (USA, Europe, etc.)

Q3: Will Social Media Scanning Content Expire?

Yes:

Reason:

Social media content changes in real-time
User attitudes evolve over time
Market trends constantly update

Recommendation:

Re-Scout after 3-6 months
Or re-scan at key points (before product launch, after major market changes)

Q4: Can Scan Historical Data?

Currently Not Supported:

Scan latest real-time content
Don't support specifying historical time periods

Alternative Solution:

If need historical trend analysis → Use traditional social listening tools
Scout focuses: Current user attitudes and latest trends

Q5: Will AI Interview/Discussion Check Social Media in Real-Time?

No:

AI Persona Answers Based On:

Persona's own background and characteristics
Won't check social media in real-time

Example:

Ask AI persona in Interview: "Would you buy ¥18 sparkling coffee?"AI Persona (Linda, price sensitive):- Answers based on persona background: "Probably not, bit expensive. I usually buy ¥12 Americano"- Won't check what others say on social mediaDifference:- Scout: Scan real users on social media- Interview: AI persona answers based on own characteristics

Final Takeaway

"Social media scanning supports 10 social media platforms (Domestic 6 + Overseas 4), scans public content. AI Interview/Discussion uses 300K+ AI persona library, doesn't depend on social media real-time data."

Remember:

✅ Scout supported platforms: Xiaohongshu, Douyin, Twitter, Instagram, TikTok
✅ Choose platforms by demographics and product type
✅ Recommend 2-3 platforms for cross-validation
✅ Only scan public content, no privacy concerns
✅ AI Interview/Discussion doesn't depend on social media, uses AI persona library
✅ Regularly re-scan (every 3-6 months)

Related Feature: Social Media Scanning Document Version: v2.1 Update Date: 2026-02-02 Update Notes: Updated terminology and platform information