Software delivery
for the AI era.
RACE Programming: Reliable Agentic Coding Excellence. Theory → Manifesto → Framework → Practice. Idea to production: 2–3 days. ~3× the output for the same budget, or the same scope for a third of the cost.
A software delivery framework, not to be confused with race conditions in concurrent programming. The name stands for Reliable Agentic Coding Excellence.
For three years we have built software with AI agents, not as an experiment but as the way we deliver. Over the last year, and especially the last six months, the tools crossed a line: an agent can now be trusted with complex production code and real analytical work.
So a new teammate joined. We call it the Silicon Software Engineer. Not a tool you point at a problem, a member of the team you direct.
That one shift breaks what came before it. Agile, Scrum, and their ceremonies exist to coordinate people. When the main producer of code is an agent, a method for managing people is managing the wrong thing.
RACE Programming is the rebuild: a delivery process re-architected around managing the Silicon Software Engineer to ship production-grade software, with verification built in as a hard gate. Everything on this site, the theory, the manifesto, the framework, and the guides, follows from that.
What are you looking for?
AI-First Theory
Why AI fundamentally changes software engineering, as a structural shift in what the bottleneck is and who owns it. Two axioms, six theorems.
Read the theory →AI-First Manifesto
What to do (and what not to do) when AI systems are co-authors of engineering work. Six values and ten principles derived from theory.
Read the manifesto →Agentic Agile
Agile's four values rewritten for AI-native delivery. If your team uses Scrum, this names precisely what breaks, what survives, and why ceremonies are now the bottleneck.
Read Agentic Agile →RACE Programming
The prescriptive SDLC framework. Pit Wall, Pit Crew, Executable User Stories, Stints: every role, artifact, and ceremony specified, not suggested.
Read the framework →Practice Guides
Step-by-step guides for your context: from Scrum, greenfield project, or existing codebase. 90-day plans with metrics.
Browse guides →AI-augmented supply-chain product
A fixed-price build: a production-shaped prototype between sales calls, a foundation-level redesign absorbed in ~10 days, shipped to production in ~3 months, with the client on a paid follow-on.
Embedded AI knowledge assistant
A two-person AI team shipped a production, document-grounded assistant across 50+ offices, at an estimated ~3× the speed and ~⅓ the cost of a conventionally-staffed team.
RACE Programming was developed and first proven at First Line Software. Organizational structure, delivery process, and technological stack built as an integrated system, not assembled from off-the-shelf parts. First to prove the economics in production. The reference implementation.
Recent writing
-
The three layers of AI-native transformation
Why organizational, process, and technological change cannot be sequenced, and why most AI adoption programs fail by trying anyway. -
Prescriptive, not permissive
Most frameworks permit variation. RACE Programming prescribes. When AI shifts engineering economics by an order of magnitude, permissive is not enough.