What is the difference between RACE Programming and AWS AI-DLC?

AWS AI-DLC (AI-Driven Development Life Cycle) is a lifecycle published by AWS in July 2025 and shipped as open source workflows (awslabs/aidlc-workflows) that run inside coding agents such as Kiro, Claude Code, Codex CLI, Cursor and GitHub Copilot. It organizes the work of AI agents into phases, stages and human approval gates. RACE Programming is a prescriptive Agentic SDLC framework that defines the human side of delivery: the Team Principal, the Pit Wall (Forward Deployed Engineer and AI Product), the Pit Crew, the Executable User Story, Kanban flow released at Pit Stops on the client's release cadence, and the share of the backlog the delivery machinery clears.

Can a RACE team use AI-DLC workflows?

Yes. In RACE Programming the Forward Deployed Engineer owns the engagement's delivery machinery: agents, workflows, guardrails and pipelines. AI-DLC workflows are one possible piece of that machinery. RACE still sets who owns the specification, who owns the acceptance tests, how releases are paced and who accepts the result.

Does AI-DLC replace sprints the same way RACE Programming does?

Both drop the two-week sprint. The original AWS post calls its short cycles Bolts, measured in hours or days. RACE Programming runs on Kanban flow: each Executable User Story is production-ready in two to three days at the floor and goes out at the next Pit Stop, a production release on the client's release cadence.

When should I choose AWS AI-DLC and when RACE Programming?

Choose AI-DLC when you want a ready, auditable agent workflow for your coding tool and your team structure already works. Choose RACE Programming when the open question is the operating model: who talks to the client, who owns the specification and the tests, how the team is staffed and paced, and how delivery scales without adding people. Many teams will use both.

Audio brief · 2 min

By Pavel Khodalev, CTO at First Line Software · October 21, 2026

Compare

RACE Programming vs AWS AI-DLC

AWS AI-DLC organizes the work of AI agents into phases, stages and approval gates, and ships that lifecycle as open source workflows for coding tools. RACE Programming prescribes the human team around the agents: who talks to the client, who owns the specification and the tests, how often software reaches production, and how delivery scales without more people.

What AI-DLC is

From a blog post to open source workflows

AWS introduced the AI-Driven Development Life Cycle in a post on the AWS DevOps blog by Raja SP, a Principal Solutions Architect at AWS, on July 31, 2025. The post describes three phases. In Inception, AI turns business intent into requirements, stories and units through Mob Elaboration, where the whole team validates the AI's questions and proposals. In Construction, AI proposes architecture, domain models, code and tests through Mob Construction. Operations follows. Sprints become Bolts, short cycles measured in hours or days, and epics become Units of Work.

The lifecycle now also ships as code. The awslabs/aidlc-workflows repository (MIT-0 license, version 2.11.1 at the time of writing) describes itself as turning AI coding assistants into structured, verifiable software-delivery workflows. Its guide lays out 5 phases and 33 stages (Initialization, Ideation, Inception, Construction, Operation) with verification gates at the phase boundaries and human approval gates at the stages. The agent roster has 14 personas: 11 domain experts (product, design, delivery, architecture, developer, quality, operations and others), 2 reviewers and an adaptive composer. Scopes such as poc, mvp, feature and classic size the ceremony to the work. The same core runs in Claude Code, Kiro, Codex CLI, Cursor, opencode and GitHub Copilot.

An earlier essay, What AWS AI-DLC gets right, and where it stops short, responds to the July 2025 post. This page compares the two as they stand today.

Sources checked October 21, 2026: the AWS blog post and the aidlc-workflows repository at version 2.11.1.

At a glance

Agent lifecycle and team operating model

Dimension	RACE Programming	AWS AI-DLC
What it is	Prescriptive Agentic SDLC framework: roles, artifacts, cadence, economics	AI-driven lifecycle published by AWS, shipped as open source agent workflows
Primary unit it designs	The human team: Team Principal, Pit Wall, Pit Crew, with the AI agent inside	The agent workflow: phases, stages, agent personas, gates
Unit of work	Executable User Story with seven components, including Gherkin acceptance and a dollar estimate	Requirements, stories and Units of Work produced in Inception
Cadence	Continuous Kanban flow; each Executable User Story production-ready in 2-3 days at the floor; Pit Stops on the client's release cadence	Bolts measured in hours or days (2025 post); stage-by-stage gates (repository)
Human checkpoints	Team Principal accepts; Pit Wall owns spec and acceptance; four Definition-of-Done gates	Human approval gates at stages; verification gates between phases
How it scales	By raising the share of the backlog the machinery clears, not by adding people	By choosing a scope that sizes the stages and ceremony to the work

What RACE does differently

It names the people and what each one owns

AI-DLC describes what the agents do and where a human approves. RACE Programming starts from the other side: the team that carries the engagement.

Roles. The Team Principal on the client side owns product direction and acceptance. The Pit Wall pairs a Forward Deployed Engineer with an AI Product. The Pit Crew is a Quality Engineer and two Software Engineers. Above several teams, a TPM coordinates plans and dependencies.
Executable User Story. The work item is a contract an agent can execute: user story, working prototype, NFR in EARS, architecture in ADR, Gherkin acceptance, test data and a delivery cost in dollars.
Independent verification. AI Product writes the Gherkin acceptance tests and owns them as the test source of truth; the Quality Engineer automates them and sets guardrails the agents cannot modify.
Flow and release cadence. Each Executable User Story is production-ready in two to three days at the floor; a Pit Stop releases it on the client's release cadence.
Machinery share of the backlog. The Forward Deployed Engineer owns the engagement's delivery machinery, and the number to watch is how much of the backlog it clears. RACE scales by raising that share, not by adding people.

Where they combine

AI-DLC as machinery inside a RACE team

The two sit at different layers, so they fit together. A RACE Forward Deployed Engineer who wants a structured, auditable agent workflow can install AI-DLC in the coding tool the team already uses and treat it as part of the engagement's machinery. Inception stages feed the Executable User Story the Pit Wall owns; Construction stages run inside the Pit Crew's guardrails; approval gates land with the role that owns the decision. The RACE Engineer can carry what works from one engagement to the next.

When to choose which

Different questions, often both

Choose AI-DLC when your team structure already works and you want a ready, documented agent workflow with gates and an audit trail in your coding tool.
Choose RACE Programming when the open question is the operating model: roles, client contact, who owns the specification and the tests, cadence, staffing and economics.
Use both when you adopt RACE and want AI-DLC as part of the delivery machinery the Forward Deployed Engineer runs.

Running AI-DLC already? The framework page shows the roles and artifacts that sit around it.

Read the framework → · RACE vs Kiro → · More comparisons →

FAQ

Frequently asked questions

What is the difference between RACE Programming and AWS AI-DLC?: AWS AI-DLC (AI-Driven Development Life Cycle) is a lifecycle published by AWS in July 2025 and shipped as open source workflows (awslabs/aidlc-workflows) that run inside coding agents such as Kiro, Claude Code, Codex CLI, Cursor and GitHub Copilot. It organizes the work of AI agents into phases, stages and human approval gates. RACE Programming is a prescriptive Agentic SDLC framework that defines the human side of delivery: the Team Principal, the Pit Wall (Forward Deployed Engineer and AI Product), the Pit Crew, the Executable User Story, Kanban flow released at Pit Stops on the client's release cadence, and the share of the backlog the delivery machinery clears.
Can a RACE team use AI-DLC workflows?: Yes. In RACE Programming the Forward Deployed Engineer owns the engagement's delivery machinery: agents, workflows, guardrails and pipelines. AI-DLC workflows are one possible piece of that machinery. RACE still sets who owns the specification, who owns the acceptance tests, how releases are paced and who accepts the result.
Does AI-DLC replace sprints the same way RACE Programming does?: Both drop the two-week sprint. The original AWS post calls its short cycles Bolts, measured in hours or days. RACE Programming runs on Kanban flow: each Executable User Story is production-ready in two to three days at the floor and goes out at the next Pit Stop, a production release on the client's release cadence.
When should I choose AWS AI-DLC and when RACE Programming?: Choose AI-DLC when you want a ready, auditable agent workflow for your coding tool and your team structure already works. Choose RACE Programming when the open question is the operating model: who talks to the client, who owns the specification and the tests, how the team is staffed and paced, and how delivery scales without adding people. Many teams will use both.

Developed in the open

Help develop this

RACE Programming is a working framework, not a final answer. If this was useful, I would like your feedback: what you think is right, what you think is wrong, and what you would change. Disagree with any part, send a better version, or use it in your own work and tell me how it went. It improves faster when people develop it together.

Write to me: paul@raceprogramming.com