09 Jan 2026

1h 3m

Spec-Driven Development: Agentic Coding at FAANG Scale and Quality — Al Harris, Amazon Kiro

AI Engineer

The podcast centers on Spec-Driven Development using Kiro, an agentic IDE, and how it enhances AI development. It emphasizes scaling AI development for complex problems, improving control over AI agents, and ensuring code quality. Spec-Driven Development is presented as a holistic SDLC approach, respecting established software development practices like Waterfall and Agile. Kiro transforms prompts into clear requirements with acceptance criteria using the EARS format, which translates into system properties for property-based testing. The integration of MCPs (Memory Context Providers) is highlighted to enhance specs across requirements generation, design, and implementation. Customizing artifacts with UI mocks and including test cases in tasks are also discussed to refine the Spec-Driven Dev toolkit.

Outlines

Part 1: Introduction to Kiro and Spec-Driven Development

Part 2: Core Methodology and Technical Framework

Part 3: Features and MCP Integration

Part 4: Customization and Process Iteration

Part 5: Live Demo and Practical Application

Part 6: System Architecture and Reproducibility

Part 7: Execution and Codebase Management

Part 8: Session Management and Workflow Organization

Part 9: Ecosystem, Performance, and Final Insights

Open full episode in Podwise

Spec-Driven Development: Agentic Coding at FAANG Scale and Quality — Al Harris, Amazon Kiro

AI Engineer

Part 1: Introduction to Kiro and Spec-Driven Development

00:20Kiro: An Agentic IDE Designed to Improve Software Development Lifecycle

Kiro: An Agentic IDE Designed to Improve Software Development Lifecycle

00:52Scaling AI Development with Spec-Driven Development and Kiro

Scaling AI Development with Spec-Driven Development and Kiro

01:32Spec-Driven Development: Respecting the History of Software Development Practices

Spec-Driven Development: Respecting the History of Software Development Practices

02:05Iterating on Ideas and Synthesizing Output in Software Development

Iterating on Ideas and Synthesizing Output in Software Development

Part 2: Core Methodology and Technical Framework

02:49Structuring Software Development Flow with Spec-Driven Development

Structuring Software Development Flow with Spec-Driven Development

03:41Property-Based Testing: Translating EARS Requirements into System Properties

Property-Based Testing: Translating EARS Requirements into System Properties

04:15Ensuring Reliable Software Delivery Through Structured Requirements and Property Testing

Ensuring Reliable Software Delivery Through Structured Requirements and Property Testing

05:32Spec as a Representation of System State, Workflow, and Tooling

Spec as a Representation of System State, Workflow, and Tooling

06:15Requirements Verification: Scanning for Ambiguity and Conflicting Constraints

Requirements Verification: Scanning for Ambiguity and Conflicting Constraints

Part 3: Features and MCP Integration

06:36Kiro Features: Steering, MCP Integration, and Software Hooks

Kiro Features: Steering, MCP Integration, and Software Hooks

07:11Sharpening the AI Toolchain: Questions and Concrete Details

Sharpening the AI Toolchain: Questions and Concrete Details

07:38Integrating MCPs into Spec-Driven Development Workflow

Integrating MCPs into Spec-Driven Development Workflow

08:34Setting Up and Managing MCPs in Kiro

Setting Up and Managing MCPs in Kiro

09:16Using MCPs for Spec Generation: Integrating with Asana

Using MCPs for Spec Generation: Integrating with Asana

10:24Generating Requirements from Asana Tasks and Code Synthesis

Generating Requirements from Asana Tasks and Code Synthesis

11:34Extending MCPs: Examples with Nobel Prize and Lofty Views APIs

Extending MCPs: Examples with Nobel Prize and Lofty Views APIs

12:30Using MCP Servers to Enhance Spec Generation Reliability

Using MCP Servers to Enhance Spec Generation Reliability

13:14Leveraging the Fetch MCP for Spec Generation

Leveraging the Fetch MCP for Spec Generation

Part 4: Customization and Process Iteration

14:20Customizing Artifacts Produced in Spec-Driven Development

Customizing Artifacts Produced in Spec-Driven Development

15:16Adding UI Mocks to Specs: Using Wireframes in Natural Language

Adding UI Mocks to Specs: Using Wireframes in Natural Language

16:05Customizing Artifacts: Adding Wireframe Diagrams and ASCII for Screens

Customizing Artifacts: Adding Wireframe Diagrams and ASCII for Screens

17:34Including Test Cases in Definitions and Tasks

Including Test Cases in Definitions and Tasks

18:06Customizing the Process: Tweaking Capabilities and Primitives

Customizing the Process: Tweaking Capabilities and Primitives

19:08Iterating on the Process: Seeking Alternatives and Avoiding Rigid Workflows

Iterating on the Process: Seeking Alternatives and Avoiding Rigid Workflows

20:13Exploring Alternatives: Using MCP Tools for Recommendations

Exploring Alternatives: Using MCP Tools for Recommendations

Part 5: Live Demo and Practical Application

21:01Live Demo: Specs in Kiro and Fine-Tuning

Live Demo: Specs in Kiro and Fine-Tuning

21:25Setting Up Agent Core with CDK and Vibe Coding

Setting Up Agent Core with CDK and Vibe Coding

22:29Adding MCP Servers: AWS Documentation and Fetch

Adding MCP Servers: AWS Documentation and Fetch

23:21Spec-Driven Development: Adding Memory and Persistence to an Agent

Spec-Driven Development: Adding Memory and Persistence to an Agent

24:35Challenging the Agent: Researching the Best Way to Implement Persistence

Challenging the Agent: Researching the Best Way to Implement Persistence

25:30Agent Research: LangGraph and Knowledge of Persistence

Agent Research: LangGraph and Knowledge of Persistence

26:09Iterating Requirements: Integrating with Checkpointing

Iterating Requirements: Integrating with Checkpointing

Part 6: System Architecture and Reproducibility

27:00EARS Requirements: Structured Natural Language for Deterministic Results

EARS Requirements: Structured Natural Language for Deterministic Results

28:32Formalizing Requirements: Correctness Properties and Property-Based Testing

Formalizing Requirements: Correctness Properties and Property-Based Testing

29:34Kiro's Spec-Driven Development: A Structured System for Reproducible Results

Kiro's Spec-Driven Development: A Structured System for Reproducible Results

30:31Mutating Previous Specs: A Living Documentation of the System

Mutating Previous Specs: A Living Documentation of the System