A practical method for AI-native software engineering. One spec. Two agents. Five artefacts. Complete traceability from intent to verification.
Spec Driven Development has roots in formal specification, design-by-contract, and the shift toward treating specs as executable artefacts. This session introduces a practical, open method — the provenance chain, the builder-tester separation, and the templates to implement it today.
90% of developers are stuck at Level 0–2 on Shapiro's maturity model — using AI as a faster tab key. They know there's a higher level but they don't have a method to get there. SDD is the method. Write a spec, delegate to agents, verify through provenance.
The unique contribution of this session is provenance — how agents communicate through documentation, how reasoning is captured as a byproduct of building, and how that creates compliance-ready evidence without additional effort.
This session is for the people who need to go back to their desk on Monday and start working differently. It gives them a method, three templates, and a workflow they can implement with tools they already have.
Writing code alongside AI assistants and feeling diminishing returns. This session gives them a structured method to move from coding with AI to directing AI — spec-first, provenance-driven, verifiable.
Building the internal infrastructure that AI-native development depends on. This session shows them how SDD integrates with CI/CD, how provenance fits into existing toolchains, and why the .sdd directory structure matters.
Responsible for quality and delivery in teams adopting AI tools. This session gives them a framework for separation of concerns, audit-ready documentation, and a process that doesn't depend on individual tool choices.
The SDD architecture: spec as the authority, provenance at the crossing, scenarios as evidence. Two agent roles communicating through documents, not orchestration frameworks. A workflow they can draw on a napkin.
Code is self-describing — any agent can read it. Provenance answers the questions code can't answer about itself: why is it this way? What was assumed? Where was the spec silent? The reasoning record that makes AI-built software navigable.
Why the agent that builds the software cannot be the agent that verifies it. Same principle as code review, enforced structurally through role definitions in the spec itself. Prose scenarios before test code.
The spec template with agent roles. The provenance template with builder and tester layers. The scenario template with prose-first verification. Markdown files. No SDK. No framework. No vendor lock-in.
How SDD's provenance chain satisfies SOC 2 change management, ISO 27001 secure development requirements, and EU AI Act documentation obligations — produced automatically as part of the build, not retrofitted.
What SDD does and doesn't give you. How it relates to LangGraph, CrewAI, Bloom, MCP, and other agent technologies. Why starting with the spec sharpens the thinking that makes every other tool more effective.
Where most developers actually sit on Shapiro's five levels. Why 90% are stuck at L0–2 and what it takes to move up. SDD as the practical method to reach L3–4.
The bottleneck has moved from implementation speed to specification quality. What a spec contains in SDD — requirements, architecture, agent roles, and process — all in one document.
The SDD architecture in detail. The builder agent, the testing agent, provenance as the communication protocol between them. Code as canonical context. The five artefacts and their purposes.
Deep dive into provenance as the core innovation. Prose scenarios before test code. How failing tests become work orders, not bug reports. The fix cycle and the layered provenance document.
SOC 2, ISO 27001, and the EU AI Act. How provenance satisfies documentation and traceability requirements as a byproduct of building. Your auditor's AI is going to ask how this was built.
15 minutes structured Q&A. Kevin works with these tools daily and handles implementation questions, edge cases, and "how does this work with my stack" directly.
The Dark Factory is the strategy — the gap between AI-native software teams and everyone else, and what it takes to cross it. SDD is the execution — the practical method that makes the transition real. One sells the problem. The other sells the discipline. Together they tell the full story.
VIEW THE DARK FACTORY SESSION BRIEF →Kevin Ryan & Associates — AI-Native · Platform Engineering · Author
"I used to direct teams of software engineers. Now I coordinate AI agents."
30 years in enterprise technology. 14 professional certifications including GitLab ×9 and GitHub ×4. 40+ enterprise clients and £20m+ in programme budgets delivered. Currently writing Spec Driven Development: AI-Native Software Engineering (sddbook.com) — the book that formalises the methodology this talk introduces. Published author of AI Immigrants. Remote-first. Budapest · Dublin · London.
If this sounds right for your audience, get in touch. We'll talk through the fit, the format, and what works for your context.