Quick Reference

AIDLC Cheatsheet

Quick reference guide for all five AIDLC phases. Keep this handy as you implement AI-Native Software Delivery.

A

Analyze

Phase 01

AI-assisted requirements gathering and research synthesis

Key Activities

  • Requirements extraction
  • Research synthesis
  • Competitive analysis
  • Feasibility assessment

Example Prompts

  • "Summarize these requirements..."
  • "What are the risks of..."
  • "Compare approaches for..."
  • "What questions should I ask..."
I

Ideate

Phase 02

AI-driven architecture design and documentation generation

Key Activities

  • Architecture diagrams
  • Design pattern selection
  • API contract design
  • Technical specifications

Example Prompts

  • "Design an architecture for..."
  • "Generate a Mermaid diagram..."
  • "What patterns fit this..."
  • "Write a technical spec for..."
D

Develop

Phase 03

AI coding agents and intelligent code generation

Key Activities

  • Delegate scoped tasks to agents
  • Review agent diffs
  • Refactoring assistance
  • Bug identification

Example Prompts

  • "Implement this function..."
  • "Review this code for..."
  • "Refactor to use..."
  • "Why is this failing..."
L

Launch

Phase 04

AI-generated tests and deployment automation

Key Activities

  • Test generation
  • CI/CD configuration
  • Deployment scripts
  • Release validation

Example Prompts

  • "Generate tests for..."
  • "Create a CI pipeline..."
  • "Write deployment script..."
  • "What should I test before..."
C

Curate

Phase 05

AI-powered monitoring and continuous improvement

Key Activities

  • Log analysis
  • Root cause analysis
  • Performance optimization
  • Documentation updates

Example Prompts

  • "Analyze these logs..."
  • "What caused this error..."
  • "How can I optimize..."
  • "Update docs to reflect..."

At a Glance

Phase Focus AI Role Output
Analyze Requirements Research assistant Specs, analysis docs
Ideate Design Architect partner Diagrams, specs
Develop Code Coding agent Working code
Launch Deploy QA engineer Tests, pipelines
Curate Maintain SRE assistant Fixes, updates

Ten Rules for AI in the SDLC

The operating model in ten sentences.

  1. AI in the SDLC is an operating model change, not a tooling purchase.
  2. Authoring is cheap. Verification is the constraint.
  3. Individual output is not organizational throughput.
  4. Governance before scale.
  5. Baseline before adoption.
  6. Measure outcomes, not activity.
  7. Humans own merges.
  8. The bar for AI-authored code is the bar for human code, applied with more suspicion.
  9. Agents are services.
  10. The compounding asset is your context, not the model.

Do's

  • Provide context in every prompt
  • Review all AI-generated code
  • Iterate on prompts for better results
  • Document AI-assisted decisions
  • Start with one phase, expand gradually

Don'ts

  • Blindly accept AI output
  • Skip security reviews
  • Expect AI to understand without context
  • Replace human judgment entirely
  • Try to implement all phases at once

What Good Looks Like

Signs your team is getting AIDLC right - qualities to aim for, not benchmark promises.

Less time on boilerplate

More of your day goes to design and review, less to repetitive typing.

Tests written by default

Coverage stops being the thing that gets cut when deadlines loom.

Docs that stay current

Documentation tracks the code instead of rotting behind it.

Humans own the review

Every agent diff is read and understood before it ships.

Frequently Asked Questions

Yes! The cheatsheet is designed to be printer-friendly. Use your browser's print function to create a physical reference you can keep at your desk.

Each phase has specific prompts: Analyze uses research and requirements prompts, Ideate uses architecture and design prompts, Develop uses code generation prompts, Launch uses testing and deployment prompts, and Curate uses monitoring and documentation prompts.

Want to Go Deeper?

Explore detailed guides for each phase of the AIDLC framework.