atypica.AI

Plan Mode: Let AI Think Before Acting

One-line summary: Plan Mode is the intent clarification layer where AI automatically generates a complete research plan with methods and steps, which you review and confirm before execution begins.

Why Do You Need Plan Mode?

The Problem with Traditional AI

Direct execution, black box operation:

  • User: "I want to understand Gen Z's attitudes toward sustainable fashion"
  • AI: Starts executing immediately (user doesn't know what it will do)
  • Result: Might be going in the wrong direction, wasting time and budget

Plan Mode's Solution

Show the plan first, then execute after confirmation:

  • User: "I want to understand Gen Z's attitudes toward sustainable fashion"
  • AI: Generates a 5-step research plan
    1. Define target user (Gen Z, 18-25 years old)
    2. Monitor social media (Little Red Book/Bilibili discussions)
    3. Group discussion (8 people discussing sustainable fashion)
    4. In-depth interviews (5 people exploring purchase intent)
    5. Generate research report
  • User: Reviews plan → Confirms or modifies
  • AI: Begins execution

Core Value

1. Transparency: See AI's Thinking Process

Users can see:

  • Which demographics AI will research
  • What research methods will be used
  • How many steps are involved
  • What outputs to expect

Benefits:

  • No more black box
  • Catch problems in advance
  • Avoid wrong direction

2. Controllability: You Can Modify the Plan

Flexible adjustments:

  • Don't like the persona selection → Modify screening criteria
  • Don't need a certain step → Delete it
  • Want deeper research → Add more interviews

vs. Traditional AI:

  • Traditional: Discover problems after 8 hours of execution, wasting resources
  • Plan Mode: Adjust before execution, zero-cost corrections

3. Collaboration: Team Can Review Together

Team usage scenarios:

  • Researcher submits requirements
  • Plan Mode generates plan
  • Team members review
  • Discuss and confirm execution

Value:

  • Ensure research direction aligns with team consensus
  • Avoid disputes after execution

Real Case Study: Bubble Coffee Product Validation

User Input (Initial Request)

Problem: Vague requirement, not sure where to start

Plan Mode Generates Plan

AI automatically analyzes the requirement and generates:

Research Plan: Bubble Coffee Product Positioning Validation

Goal: Determine product positioning (health vs social vs other)

Steps:

  1. Define Target User Characteristics

    • Demographics: 25-35 years old, first-tier city professionals
    • Screening criteria: Regular coffee consumers, willing to try new products
    • Estimated participants: 8 people

  2. Market Trend Analysis

    • Use Product R&D Agent
    • Analyze coffee market trends
    • Output 3 possible positioning directions

  3. Group Discussion to Test Positioning

    • Use Discussion
    • 8 people discuss 3 positioning directions
    • Quickly identify the most popular direction

  4. In-Depth Interviews to Verify Motivation

    • Use Interview
    • 1-on-1 interviews with 5 high-intent users
    • Deep dive into why they like this positioning

  5. Generate Research Report

    • Product positioning recommendations
    • Target user persona
    • Pricing and packaging suggestions

Estimated time: 8-12 hours Expected output: Complete product positioning report

User Reviews Plan

After seeing the plan:

✅ Satisfied aspects:

  • Research steps are clear
  • The sequence of discussion before interviews makes sense

⚠️ Need adjustments:

  • "25-35 is too broad, I want to focus on 28-32-year-old women"
  • "Can we add packaging design testing?"

User modifications:

  1. Adjust target user: 28-32 years old, female professionals
  2. Add step: Upload 3 packaging design mockups, test feedback via Discussion

AI Updates Plan

Based on user feedback, AI automatically updates:

Research Plan: Bubble Coffee Product Positioning Validation (v2)

Goal: Determine product positioning + packaging design validation

Steps:

  1. Define Target User Characteristics

    • Demographics: 28-32 years old, first-tier city female professionals (modified)
    • Screening criteria: Regular coffee consumers, value product aesthetics
    • Estimated participants: 8 people

  2. Market Trend Analysis (same as above)

  3. Group Discussion to Test Positioning (same as above)

  4. In-Depth Interviews to Verify Motivation (same as above)

  5. Packaging Design Testing (new)

    • Upload 3 packaging design mockups
    • Discussion about which is more appealing
    • Output packaging recommendations

  6. Generate Research Report (updated: includes packaging recommendations)

Estimated time: 10-14 hours Expected output: Product positioning report + packaging design recommendations

User Confirms Execution

User: "The plan looks good, let's start executing"

AI: Immediately begins execution according to the confirmed plan

4 Usage Scenarios for Plan Mode

Scenario 1: First-time User, Unsure How to Submit Requirements

User dilemma:

  • Know we need to research, but don't know how to describe requirements
  • Worried requirements won't be professional enough

Plan Mode value:

  • Users only need simple description
  • AI automatically fills in research details
  • After seeing plan, users understand "that's how it can be done"

Case study:

User: "I want to know what young people think about our brand"

Plan Mode generates:

  1. Define "young people" (18-30 years old)
  2. Scout Agent monitors social media discussions
  3. Interview to understand brand perception
  4. Discussion to test new positioning
  5. Generate brand perception report

Scenario 2: Complex Project, Need to Confirm Research Path

User dilemma:

  • Research project is complex with multiple phases
  • Worried AI might misunderstand and execute in wrong direction

Plan Mode value:

  • Show complete research pathway
  • Users can identify gaps or errors in advance
  • Ensure AI understands the requirements

Case study:

User: "I want to conduct complete research from idea to validation for a new product"

Plan Mode generates 8-step plan:

  1. Market opportunity analysis
  2. Cross-disciplinary ideation
  3. Idea discussion and filtering
  4. In-depth interview validation
  5. Packaging design testing
  6. Pricing testing
  7. Channel strategy discussion
  8. Generate complete product plan

User feedback: "This is perfect and comprehensive, exactly what I need"

Scenario 3: Team Collaboration, Need to Show Plan to Colleagues

User dilemma:

  • Need to present research plan to boss or team
  • Traditional AI executes immediately, can't preview

Plan Mode value:

  • Generate visual research plan
  • Team can discuss and review together
  • Confirm before execution, avoid disputes

Case study:

Brand Manager: "We want to reposition our brand, but need to show the research plan to our boss first"

Using Plan Mode:

  1. Generate 5-step research plan
  2. Export plan as PDF
  3. Team meeting review
  4. Boss confirms execution
  5. Avoid "discovering boss's dissatisfaction after execution"

Scenario 4: Limited Budget, Need to Control Research Scope

User dilemma:

  • Budget or time is limited
  • Need to selectively execute research steps

Plan Mode value:

  • See full plan, then cut unnecessary steps
  • Execute the most important parts first
  • Save costs

Case study:

User: "I want to research user personas, but budget is limited"

Plan Mode generates:

  1. Scout Agent observation (10 rounds)
  2. Discussion validation (8 people)
  3. In-depth interviews (10 people)
  4. Generate persona report

User adjusts:

  • Remove Interview (most time-consuming)
  • Keep only Scout + Discussion
  • Save 50% time and cost

Plan Mode vs Direct Execution

DimensionPlan ModeDirect Execution
TransparencyFully transparent, see all stepsBlack box, don't know what AI is doing
ControllabilityCan modify plan, flexible adjustmentsCan't adjust, only restart
RiskCatch problems early, zero-cost correctionsProblems discovered after execution, resources wasted
CollaborationTeam can review planHard to explain to others
Use casesComplex projects, first-time use, team collaborationSimple tasks, experienced users
Time+5-10 minutes review timeImmediate execution

Recommendation: Use Plan Mode in most cases to avoid wrong direction

Frequently Asked Questions

Q1: How Much Extra Time Does Plan Mode Add?

Plan review time: 5-10 minutes Value: Avoid wrong direction, save hours to days

Calculation:

  • Without Plan Mode: Execute for 8 hours, discover wrong direction, restart = 16+ hours total
  • With Plan Mode: Review 10 minutes + execute 8 hours = 8.17 hours total

Conclusion: Plan Mode actually saves time

Q2: Can I Fully Customize the Generated Plan?

Can customize:

  • Modify research steps
  • Adjust target users
  • Add/remove research methods
  • Change number of participants and scope

Cannot customize:

  • Can't ask AI to do research methods it doesn't support
  • Basic framework limited by atypica.AI's capabilities

Q3: Can I Regenerate the Plan If I'm Not Satisfied?

Yes:

  • Click "Regenerate"
  • AI generates new plan based on requirements
  • Can regenerate multiple times

Recommendation:

  • For small adjustments, modify directly (faster)
  • For completely wrong direction, regenerate

Q4: Is Plan Mode Mandatory?

Not mandatory:

  • Experienced users can skip Plan Mode
  • Simple tasks can execute directly

But strongly recommended:

  • Use Plan Mode on first use to understand the process
  • Use Plan Mode for complex projects to avoid errors
  • Use Plan Mode for team collaboration to reach consensus

Q5: How Long After Plan Generation Do You Need to Confirm?

No time limit:

  • Plan is saved in system
  • Can review anytime
  • Can share with team for discussion before confirming

Recommendation:

  • Confirm same day to avoid losing context
  • Important projects can discuss for 1-2 days

Practical Tips

1. Must Use Plan Mode on First Use

Reasons:

  • Understand how atypica.AI works
  • Learn how to submit better requirements
  • Avoid wasting resources due to unfamiliarity

Experience:

  • 1st time: "Oh, that's how it works"
  • 3rd time: Start understanding how to submit requirements
  • 10th time: Can consider skipping Plan Mode

2. Must Use Plan Mode for Complex Projects

What's a complex project:

  • Involves 3+ research steps
  • Needs multiple research methods
  • Limited budget or time requiring precise control

Value:

  • Optimize research path in advance
  • Avoid discovering wrong direction mid-project
  • Save time and cost

3. Must Use Plan Mode for Team Collaboration

Reasons:

  • Need to explain research plan to boss/colleagues
  • Need team discussion and consensus
  • Avoid post-execution disputes

Workflow:

  1. Researcher uses Plan Mode to generate plan
  2. Export plan document
  3. Team meeting review
  4. Confirm execution after approval

4. Can Skip Plan Mode for Simple Tasks

What's a simple task:

  • Only needs 1-2 steps
  • Research direction is very clear
  • Experienced user

Examples:

  • "8 people discuss this product concept"
  • "Monitor Little Red Book for camping-related discussions"

Can execute directly, no need for Plan Mode

Document version: v2.0 | 2026-01-15 | User-focused perspective

Last updated: 1/20/2026