Elevating Software Quality, Traceability,
and Human-AI Collaboration
Haim Michael
Spec-Driven Development
with Kiro IDE
Scan Me!
Haim Michael
● 25+ years of professional experience,
based in Tel Aviv.
● Educator and trainer in academia and
high-tech (Java, Python, JS/TS, Kotlin,
PHP, Scala, Design, Architecture).
● Academic background in Economics,
Accounting, and Business Management.
● Business experience in mobile games and
applications @ jacado.com (300+ titles).
blog.lifemichael.com
AI Driven
Development
01 The State of AI Driven
Development Today
Spec-Driven
Development
02 Specifications as the
Source of Truth
The Kiro IDE
03 Kiro IDE ensures end-to-
end traceability
Agents Orchestration
04 Orchestrating AI Agents
with Human Intent
Table of Content
AI-Driven Software
Development
The State of AI-Driven Development Today
01
AI-Driven Software Development (1/2)
Code
Completion
AI automatically
completes our code
Getting
Explanations
Explanations for code
developed by others
Skeleton
Generation
Getting a skeleton we
we can work on
Agentic
Coding
We can use AI agents
doing the coders work
AI-Driven Software Development (2/2)
Code
Review
We can use AI to
review our code
Generating
Tests
We can easily instruct
AI to create the tests
Creating
Documents
We can easily use AI to
create documentation
Code
Refactoring
We can use AI to
refactor the code
Vibe Coding
Intuition-led,
exploratory coding with
minimal structure
Prompt Driven
Development
Spec-Driven
Development
The Methodology
Iterative development
driven by detailed
prompts and context
Specifications drive
design, code, tests, and
traceability
Spec-Driven Development
The Comeback of The Old New Methodology
02
Coding is no longer
a Scarce skill.
AI can generate code at scale,
refactor entire systems, and
implement features in seconds.
Clarity of intent is the one that is
more difficult to achieve.
Vibe Coding does
not Scale.
It lacks traceability and a
stable source of truth.
Prompt Driven
Amplifies The Problem
Prompts are transient, informal,
and rarely treated as durable
artifacts. The systems evolve,
intent is lost, decisions become
opaque, and the technology debt
grows.
Spec-Driven
Development
The Specs Role
for Organizations
Institutional Memory
and Governance
for Humans
Clarity &
Alignment
for AI
Constraints &
Guidance
“
‫ס‬
‫ו‬
‫ף‬
‫מ‬
‫ע‬
‫ש‬
‫ה‬
‫ב‬
‫מ‬
‫ח‬
‫ש‬
‫ב‬
‫ה‬
‫ת‬
‫ח‬
‫י‬
‫ל‬
‫ה‬
“
End of Action with Forethought
The Kiro IDE by Amazon
End to End Traceability
03
Kiro IDE by Amazon
Tasks that map to
specific requirements
VS Code
Kiro was built on top of
the well known VS Code
Agentic IDE
Autonomous AI Agents
doing the work
Traceability MCP Support
Empower the IDE with
MCP servers
Kiro IDE by Amazon
Steering Files Specs Code Ready
Requirements Design List of Tasks
The
Steering
Files
The Steering
Documents
The
Specifications
Requirements
The Easy Approach to Requirements Syntax (EARS)
The
Specifications
Design
The
Specifications
Tasks List
Agents Orchestration
Coordinating Multiple AI Agents
04
Kiro IDE AI Agents
Custom Agents
We can create agents for
specific responsibilities
Autonomous Agent
The frontier agent that
executes the tasks, etc.
Agent Hooks
Lightweight agents
triggered on events
Agent
Hooks
Tasks List
THANKS
Questions & Answers
haim.michael@lifemichael.com
+972 54 6655837
blog.lifemichael.com
e

Spec-Driven Development with Kiro: Elevating Software Quality, Traceability, and Human-AI Collaboration