How to Use AI Expert Agent to Interview AI Personas?

Question Type

User Manual Question

Quick Answer

Sage (AI Expert System) is NOT for "interviewing AI personas".

Sage's True Purpose:

✅ Create evolvable AI expert advisors (based on your professional materials)
✅ Consult experts anytime (like consulting real experts)
✅ Let experts continuously evolve through supplementary interviews

If You Want to "Interview AI Personas":

→ Use Interview feature (AI Persona Interview)
→ Not Sage

Sage vs Interview Difference:

Interview: You interview AI personas (user simulation) to understand user needs
Sage: You consult AI experts (expert simulation) for professional advice

What is Sage?

Core Positioning: Train Your AI Expert Advisor

Not:

❌ Tool for interviewing users
❌ Tool for market research

But Rather:

✅ Train an AI expert based on your professional materials
✅ Consult this expert anytime for professional advice
✅ Let expert continuously evolve through dialogue and interviews

Sage vs Interview Comparison

Dimension	Sage (AI Expert System)	Interview (AI Persona Interview)
Role	AI expert advisor	AI user simulation
Your Identity	Consultant	Interviewer/Researcher
Dialogue Purpose	Get professional advice	Understand user needs
Knowledge Source	Your uploaded professional materials	atypica's 300K+ persona library
Use Cases	Enterprise knowledge transfer, personal learning, industry research	Product testing, market research, user insights
Output	Professional advice and guidance	User feedback and insights
Evolution Method	Supplementary interviews fill knowledge gaps	No evolution (personas fixed)

Complete Sage Usage Flow

Step 1: Create AI Expert

Upload Professional Materials:

Supported Formats:

PDF, TXT, Markdown, Audio, DOCX (up to 10 files)
Text paste
URL import (auto-scrape webpage content)

Example Scenario: Enterprise UX Design Expert

Upload Knowledge Sources:1. design-system-v2.3.pdf (50-page design specifications)2. ux-principles.md (UX design principles)3. https://company.com/design-tokens (Design Token documentation)AI Auto-Processing:→ Extract plain text: 65,000 words professional knowledge→ Build Memory Document→ Generate expert bio and recommended questions

Step 2: AI Analyzes and Builds Expert

AI Auto-Analysis:

Uses Claude Sonnet 4.5 to analyze knowledge sources:1. Extract expert bio2. Identify expertise areas3. Generate recommended questions4. Build structured Markdown memory document5. Identify knowledge gapsOutput:✓ Memory Document v1 created✓ Expert bio: I'm your company's UX design expert, familiar with design system v2.3...✓ Expertise: Design system specifications, component library usage, design tokens...✓ Knowledge gaps: 3 (mobile specifications, animation specs, design decision cases)

Step 3: Consult AI Expert (Dialogue)

Like consulting a real expert:

You: "What should our button corner radius be?"Sage: "According to our design system v2.3, button corner radius uniformly uses 6px.      Specific rules:      - Primary Button: 6px corner radius + 16px padding      - Secondary Button: 6px corner radius + 12px padding      - Icon Button: 8px corner radius (circular button)      This is to form unified visual hierarchy with our card corner radius (8px)      and input box corner radius (4px)."→ Next dialogue, Sage still remembers design specifications, no need to repeat

Step 4: Supplementary Interview (Expert Evolution)

When Sage discovers knowledge gaps:

Trigger Conditions:

Discovers unanswerable questions in dialogue
Knowledge gaps identified in initial analysis
User proactively creates supplementary interview

Supplementary Interview Flow:

Sage discovers knowledge gaps:- Gap 1: Missing mobile design specifications- Gap 2: Animation specifications incomplete- Gap 3: Missing design decision casesYou create supplementary interview:→ AI interviewer generates interview plan→ AI interviewer asks you questions→ You answer (acting as expert)→ AI auto-updates knowledge base (creates Memory Document v2)→ Marks resolved Knowledge GapsInterview Example:AI: "Hello! I want to supplement some knowledge about mobile design specifications.    First question: What are the core differences between our company's iOS and Android design specs?"You: "Three main differences:    1. Navigation mode: iOS uses bottom Tab Bar, Android uses Bottom Navigation    2. Back button: iOS top left, Android uses system back key    3. Corner radius: iOS prefers larger radius (8-12px), Android 4-6px"AI: "Understood. You mentioned iOS uses bottom Tab Bar, can you detail its design specifications?    Such as height, icon size, activated state styling?"You: "Bottom Tab Bar:    - Height: 80px (including safe area)    - Icon: 24×24px, activated color is primary-500    - Text: 12px, activated font-weight: 600    - Spacing: Icon-text spacing 4px"[After interview ends, AI auto-updates knowledge base]Memory Document v2 created ✓- New content: Mobile design specifications (detailed)- Resolved Gaps: 3- Change description: Supplemented iOS/Android mobile design specifications

Sage Use Cases

Scenario 1: Enterprise Knowledge Transfer

Background: Senior employee about to retire, needs to transfer 20 years experience

Operations:

Step 1: Create "Senior UX Designer - Zhang Wei" SageUpload:- Project docs (50+ projects)- Design specifications (company design system v1-v3)- Presentation PPTs (20+ design sharing session PPTs)- Design review recordsStep 2: AI builds expert knowledge baseStep 3: New employees consult SageNew hire: "What are our product table design best practices?"Sage (Zhang Wei): "Based on my 20 years experience, B2B product table design                  should follow these principles:                  1. Clear information hierarchy...                  2. Convenient operations...                  3. Performance optimization...                  I used this approach in projects X and Y..."Value:- Knowledge permanently saved (designer retires, knowledge remains usable)- Consult anytime (new employees not limited by senior employee's time)- Continuous evolution (new employees' questions continuously optimize Sage)

Scenario 2: Personal Learning Assistant

Background: University student studying machine learning course

Operations:

Step 1: Create "Machine Learning Tutor" SageUpload:- Course PPTs (40 lectures)- Course notes (Markdown)- Textbook PDF (Deep Learning Book)- Papers (BERT, Transformer, etc.)- Assignment code (Jupyter Notebooks)Step 2: Consult during learningStudent: "How does backpropagation algorithm work?"Sage: "According to our course lecture 5, backpropagation works like this:      1. Forward propagation calculates output...      2. Calculate loss function...      3. Backpropagation calculates gradients...      You already implemented this algorithm in assignment 3 (see hw3_backprop.ipynb).      Want me to explain a specific step?"Value:- Personalized learning (based on own notes)- Timely Q&A (don't need to wait for office hours)- Fill knowledge gaps (Sage proactively discovers learning gaps)

Scenario 3: Industry Research Expert

Background: FinTech industry analyst tracking industry trends

Operations:

Step 1: Create "FinTech Industry Expert" SageUpload:- Industry reports (McKinsey, iResearch, Gartner)- Company announcements (10+ FinTech company earnings)- Research papers (blockchain, payments, digital currency)- News articles (50+ articles)Step 2: Quick insightsAnalyst: "What are core FinTech trends for 2025?"Sage: "Based on integrating 5 industry reports and 20+ company announcements,      2025 FinTech has 3 core trends:      1. AI-driven risk control upgrade...      2. Embedded finance explosion...      3. Digital currency compliance...      Want me to deeply analyze a specific trend?"Value:- Quickly integrate multiple reports- Continuous updates (immediate updates when new reports release)- Discover contradictions (auto-discover viewpoint differences in different reports)

Sage vs NotebookLM

Core Differences

Dimension	Sage	Google NotebookLM
Core Positioning	Evolvable AI expert agent system	AI research assistant and knowledge management tool
Core Value	Build continuously learning professional AI advisor	Quickly understand and utilize existing documents
Knowledge Management	Versioned memory documents + active learning	Static analysis based on uploaded content
Learning Mechanism	Proactively identify knowledge gaps and supplement through interviews	Passively respond to user queries
Unique Features	Knowledge Gap tracking, supplementary interviews, knowledge evolution	Audio Overview (documents to podcast), Deep Research

Sage Emphasizes "Training Expert":

Knowledge is dynamically evolving
System proactively discovers knowledge gaps
Through structured interviews complete knowledge
Complete knowledge evolution tracking

NotebookLM Emphasizes "Understanding Materials":

Knowledge is static (based on uploaded documents)
Passively responds to user queries
Innovative output formats (podcasts, video summaries)
Quick start, low barrier to use

Correct Tool for "Interviewing AI Personas": Interview

If you want to "interview AI personas" to understand user needs, should use Interview feature:

Interview Feature Introduction

Core Purpose:

Interview AI personas (user simulation)
Understand user needs, pain points, decision process
Test product concepts, pricing, packaging, etc.

Usage Flow:

Step 1: Create Interview projectDescription: "Test acceptance of sparkling coffee new product"Step 2: Select AI personas (from 300K+ persona library)- Linda (28, Product Manager, price sensitive)- Emma (32, Accountant, health anxious)- Chloe (26, Designer, social early adopter)Step 3: Start interviewYou: "Would you buy ¥18 sparkling coffee?"Linda: "Probably not, bit expensive. I usually buy ¥12 Americano."You: "Why do you think it's expensive?"Linda: "I usually buy ¥12 Americano, ¥18 exceeds my budget."You: "If it's ¥12, would you buy?"Linda: "Still no, mainly because sparkling + coffee sounds weird..."→ Deeply dig user needs and pain points

Difference:

Sage: Consult expert for professional advice
Interview: Interview users to understand user needs

FAQ

Q1: Can Sage Be Used for Market Research?

No, Sage is not a market research tool:

Sage's Purpose:

✅ Create expert advisor (based on your professional materials)
✅ Consult expert advice
❌ Cannot do market research

For Market Research Use:

Scout (Social media scanning)
Interview/Discussion (Interview AI personas)
Plan Mode (Comprehensive research)

Q2: Can Sage Interview Real Users?

No:

Sage's Role:

AI expert advisor (you consult it)
Not an interview tool

For Interviewing Real Users Use:

AI Persona Interview (interview AI personas)
Or offline real person interviews

Q3: What Does Sage's "Supplementary Interview" Mean?

Supplementary Interview Purpose: Fill Sage's Knowledge Gaps

Not:

❌ Sage interviews users
❌ Sage interviews other AI

But Rather:

✅ AI interviewer asks you questions
✅ You answer (acting as expert)
✅ Supplement Sage's missing knowledge

Flow:

Sage discovers knowledge gap:"My knowledge base lacks mobile design specifications"You create supplementary interview:AI interviewer: "What are our iOS design specifications?"You: "iOS design specifications include the following..."After interview ends, Sage auto-updates knowledge base

Q4: What's the Difference Between Sage and Memory System?

Completely different features:

Memory System:

Remembers your research needs and history
Personalized AI recommendations
Auto-extract and organize

Sage:

Create AI expert advisor
Based on your uploaded professional materials
Consult expert advice anytime

Relationship:

Memory: User-level memory
Sage: Expert-level memory
Dual-layer architecture, complementary

Q5: What Scenarios Suit Sage?

Suitable:

✅ Enterprise knowledge transfer (senior employee experience)
✅ Personal learning assistant (course notes organization)
✅ Industry research expert (continuous industry tracking)
✅ Professional consulting (legal, medical, technical fields)

Not Suitable:

❌ Market research (use Scout/Interview)
❌ User interviews (use Interview)
❌ Real-time data queries (like stock prices)
❌ General Q&A (ChatGPT more suitable)

Final Takeaway

"Sage is not for 'interviewing AI personas', but for creating evolvable AI expert advisors. If want to interview AI personas (user simulation), should use Interview feature."

Remember:

✅ Sage purpose: Create AI expert → Consult expert advice → Supplement knowledge for expert evolution
✅ Interview purpose: Select AI personas → Interview users → Understand user needs
✅ Completely different: Sage = Expert advisor, Interview = User interview
✅ Sage suitable: Enterprise knowledge transfer, personal learning, industry research
✅ Interview suitable: Product testing, market research, user insights

Related Feature: Sage System, AI Persona Interview Document Version: v2.1