05 — Planning: PRD & AI Baseline

Purpose: Create a comprehensive Product Requirements Document (PRD) and plan AI Baseline integration for your MVP

Outcome: Have a complete PRD that guides implementation and a well-planned AI Baseline integration (if applicable)

Audience: PM / Dev / Both

Time: 2-4 hours

Prerequisites: 05 — Planning: User Stories - Finalized and prioritized user stories

Learning Outcomes

By the end of this chapter, you will be able to:

Use AI tools to generate a comprehensive PRD from user stories
Review and refine PRD sections with AI assistance
Ensure all PRD sections are complete and clear
Plan AI Baseline integration including service selection, model choice, and prompt design
Document AI features and integration points in the PRD

Jobs-to-Be-Done

When: I have finalized user stories and need to create a detailed PRD
I want: To generate a comprehensive PRD and plan AI Baseline integration
So that: I have a clear blueprint that guides implementation including AI components

Inputs

Finalized user stories from 05 — Planning: User Stories
Product context and value proposition
Understanding of target users and problems
Access to AI tools (ChatGPT, Claude, or Cursor)
PRD Template for reference

Activities

1. Prepare Materials for PRD Generation

Before generating the PRD, gather everything you'll provide to AI:

What You Need:

Finalized and prioritized user stories
Product description and value proposition
Target user personas
Key problems being solved
Innovation aspects
MVP scope boundaries

Context to Provide:

Complete list of user stories
Product vision and purpose
Target users and personas
Core problems and solutions
MVP scope (in-scope vs. out-of-scope)

2. Generate Product Requirements Document (PRD)

Use AI tools to generate a comprehensive PRD based on your finalized user stories.

Process:

Open your AI tool (ChatGPT, Claude, or Cursor)
Provide your finalized user stories and product context
Reference the PRD Template structure
Request generation of a PRD following the template structure
Specify that all sections should be complete and clear
Request specific acceptance criteria for each feature

PRD Structure (see template):

Purpose & Vision
Scope (Release 1)
Key Personas
User Journeys
Functional Requirements
Non-Functional Requirements
Information Architecture & Data Model
UX References
Acceptance Criteria
Analytics & Success Metrics
Content & States
Accessibility & Design System
Security, Privacy, Compliance
Glossary

Example Prompt:

Based on these user stories: [list user stories],
Generate a comprehensive PRD for MVP Release 1 following this structure:
[reference PRD template structure]

Include:
- Purpose & Vision
- Scope (Release 1) with in-scope and out-of-scope features
- Key Personas
- User Journeys for each major flow
- Functional Requirements with acceptance criteria
- Non-Functional Requirements
- Information Architecture & Data Model
- UX References (mockups will be added later)
- Acceptance Criteria for each feature
- Analytics & Success Metrics
- Content & States (empty states, error states)
- Accessibility & Design System requirements
- Security, Privacy, Compliance requirements
- Glossary of key terms

Make sure all sections are complete and clear.

3. Review and Refine PRD

Review Process:

Review the generated PRD section by section
Check for completeness and clarity
Verify alignment with user stories
Ensure acceptance criteria are specific and measurable
Refine sections that need improvement

Refinement Process:

Identify sections that need improvement
Use AI to refine specific sections
Add missing details or requirements
Ensure consistency across sections
Verify all user stories are addressed

Example Refinement Prompt:

Here's my PRD: [PRD content]

Please:
1. Review the Functional Requirements section
2. Ensure all user stories are addressed
3. Add missing acceptance criteria
4. Make acceptance criteria more specific and measurable
5. Ensure consistency with User Journeys section

📝 Note: The PRD serves as the blueprint for your MVP. Spend time getting it right - it will guide all implementation decisions.

4. Plan AI Baseline Integration

If your MVP includes AI functionality, plan the AI Baseline integration during the Planning phase. This ensures AI components are designed alongside the rest of your product architecture.

Process:

Step 1: Select AI Service

Choose your AI provider:

OpenAI: GPT models, embeddings
Anthropic: Claude models (recommended for this stack)
Custom: Build your own (not recommended for MVP)

Selection Criteria:

Quality of responses for your use case
Cost considerations
API reliability and rate limits
Integration complexity

Step 2: Choose Model

Select appropriate model for MVP:

Claude 3.5 Sonnet: Best quality, balanced cost
Claude 3 Haiku: Fast responses, lower cost
GPT-4: Alternative option
GPT-3.5: Lower cost option

Model Selection Criteria:

Required response quality
Response speed needs
Cost constraints
Token limits for your use case

Step 3: Design AI Integration

Define AI features and integration points:

Questions to Answer:

What will AI do in your product?
What inputs does it need?
What outputs does it produce?
How will users interact with AI features?
Where in the user journey does AI add value?

Integration Points to Define:

Where AI is triggered (user action, background process, scheduled)
What data flows to AI (user input, context, history)
What AI produces (responses, insights, recommendations)
How users see AI output (inline, modal, notification)

Step 4: Design Prompt Templates

Create prompt structures for AI interactions:

System Prompts:

Define AI behavior and role
Set context and constraints
Specify output format requirements
Define error handling approach

User Prompt Templates:

Structure user input for AI
Include necessary context
Format data for AI processing
Handle variable inputs

Response Formatting:

Define expected output structure
Specify JSON or text format
Define parsing requirements
Handle edge cases

Error Handling and Fallbacks:

What happens if AI fails?
Fallback mechanisms
User-friendly error messages
Retry strategies

Step 5: Document API Requirements

Define technical requirements:

API Configuration:

API keys and authentication
Rate limiting considerations
Error handling strategy
Response processing requirements
Token usage and limits

Integration Requirements:

Where API calls are made (frontend, backend, functions)
Data flow and processing
Caching strategies
Security considerations

Step 6: Include AI in PRD

Ensure AI Baseline is documented in your PRD:

PRD Sections to Update:

Functional Requirements: Add AI features with acceptance criteria
User Journeys: Include AI integration points in flows
Acceptance Criteria: Define criteria for AI features
Technical Requirements: Document AI integration requirements
Data Model: Include AI-related data structures

Example (Retrospective Tool):

AI Service: Anthropic Claude 3.5 Sonnet
Use Case: Generate insights from retrospective feedback
Input: User feedback from retrospective session
Output: Structured insights with themes, action items, and recognition points
Integration Point: After user submits feedback, AI generates insights
Prompt Design: System prompt defines retrospective facilitator role, user prompt includes feedback data

Example Prompt Structure:

System: You are a retrospective facilitator. Analyze feedback and provide structured insights.
User: [Retrospective feedback data]
Output Format: JSON with themes, action items, recognition

5. Finalize PRD

Final Review Checklist:

All PRD sections are complete
User stories are addressed in Functional Requirements
Acceptance criteria are specific and measurable
User Journeys are complete and include AI integration points (if applicable)
AI Baseline planning is documented (if applicable)
PRD is ready for mockup creation and implementation

Documentation Format:

Use the PRD template structure
Save as a markdown or document file
Include all sections
Ensure formatting is clear

💡 Tip: Start simple with AI. You can enhance AI features later. Focus on core value delivery first. 📝 Note: AI Baseline planning happens here in Planning. Implementation will follow in Build Features where Cursor will build the AI components as part of the comprehensive plan. ⚠️ Warning: Don't skip AI planning if your product relies on AI. Proper planning ensures AI integration aligns with your product vision.

Apply It Now

Task: Create your PRD and plan AI Baseline integration

Prepare user stories and product context
Use AI to generate comprehensive PRD following the template structure
Review and refine PRD sections
Plan AI Baseline integration (if applicable):
- Select AI service and model
- Design AI integration points
- Design prompt templates
- Document API requirements
Include AI features in PRD (if applicable)
Finalize PRD document

Artifacts: You'll create:

Comprehensive PRD document
AI Baseline plan/design (if applicable)
Documentation of AI integration points

Artifacts

You'll create:

Product Requirements Document (PRD)
AI Baseline plan/design (if applicable)
Documentation of AI service, model, and prompt design

Worked Example

Situation: Building a B2B AI SaaS retrospective tool

PRD Process:

Preparation: Gathered 8 finalized user stories, product context, and personas
PRD Generation: Used Claude to generate comprehensive PRD
- Provided user stories and PRD template structure
- Generated all sections: purpose, scope, personas, journeys, requirements
- Included acceptance criteria for each feature
- Reviewed and refined with AI assistance
AI Baseline Planning: Planned AI integration
- AI Service: Selected Anthropic Claude 3.5 Sonnet
- Use Case: Generate insights from retrospective feedback
- Integration: After user submits feedback, AI generates insights
- Prompt Design: System prompt defines retrospective facilitator role
- Output: JSON with themes, action items, recognition points
PRD Integration: Added AI features to PRD
- Added AI features to Functional Requirements
- Included AI integration points in User Journeys
- Defined acceptance criteria for AI features
- Documented technical requirements for AI integration
Finalization: Reviewed complete PRD, ensured all sections complete

Time: 3 hours total (2 hours PRD, 1 hour AI Baseline planning)

Checklist

Before proceeding to mockup creation, verify:

PRD generated using AI tools
PRD reviewed and refined
All PRD sections completed (Purpose, Scope, Personas, User Journeys, Requirements, etc.)
PRD includes acceptance criteria for each feature
PRD finalized and ready for implementation
AI Baseline planned and designed (if applicable)
AI features documented in PRD (if applicable)
AI integration points defined in User Journeys (if applicable)

Self-Assessment

Exit Criteria

You're ready to proceed to mockup creation when:

PRD is complete and reviewed
All PRD sections are filled out
Acceptance criteria are defined for each feature
AI Baseline planning is complete (if applicable)
AI features are documented in PRD (if applicable)
PRD is ready for mockup creation and implementation

Dependencies & Next Steps

Prerequisites Completed

05 — Planning: User Stories - Finalized and prioritized user stories

Next Steps

Proceed to 05 — Planning: Mockups to create visual mockups based on your PRD

What This Enables

Completing PRD and AI Baseline planning enables:

Clear blueprint for implementation
Structured approach to building
Well-planned AI integration
Better understanding of scope and requirements

PRD Template - Template for creating your PRD
05 — Planning - Return to Planning overview

💡 Tip: Don't rush the PRD phase. A solid PRD will save significant time during development. 📝 Note: The PRD serves as the blueprint for your MVP. Spend time getting it right. ⚠️ Warning: Don't skip AI planning if your product relies on AI. Proper planning ensures AI integration aligns with your product vision.

Previous05 — Planning: User Stories Next05 — Planning: Mockups

Last updated 1 month ago