The Soil Beneath the Code
Published
What engineering teams can learn from farmers about sustainable systems.
Scott Scoble is a builder, writer, and speaker focused on the human side of AI systems. He's the author of The Soil Beneath the Code and A Walk in Mentoring, and creator of Speckl — a compile-time provenance layer for regulated industries. He advises engineering leaders on how to build AI systems that cultivate human capability instead of strip-mining it.
Diagnose before prescribing. Know what's actually depleted before you fertilize.
Cultivate capability, don't extract output. Build systems that get stronger with use.
Hold the line on human infrastructure. Dark provenance for every agent output.
Published
What engineering teams can learn from farmers about sustainable systems.
Published
Three questions that change how people learn.
Coming 2026
Formal verification for the AI age. How to audit what your AI actually did.
A state machine specification language compiler that produces auditable code with embedded provenance. Compiles SpeckDL to TypeScript, Z3 SMT verification, and compliance artifacts (PROV-O, CycloneDX, SPDX). Fulfills NIST SA-11.
Agentic engineering framework — tools, workflows, and skills for building with AI agents. Patterns for subagent dispatch, skill-based development, and provenance tracking.
Workshop presentation on using Skills and Subagents to guide AI-assisted development, demonstrated through practical architecture decisions. Built with Reveal.js and Nix flakes.