Agentic Software Engineering — The Future of Code publishes a practical guide arguing agents shift engineering from coding to architecture, emphasizing intent, risk boundaries, and evidence for trustworthy systems. This reframes where teams invest effort — more in orchestration, intent-design, and safety tooling than in line-by-line coding, aligning with Principle 01 and Principle 14 for outcome engineers.

Red/green TDD lays out a testing-first pattern for coding agents that forces agents to produce failing tests before implementing code. Applying this pattern reduces regressions and unused code when agents generate bits of systems, making automated output auditable and testable (Principle 14, Principle 16).

The Claude C Compiler: What It Reveals About the Future of Software demonstrates that AI can assemble a competent compiler, shifting software work toward design, orchestration, and stewardship while surfacing hard IP and governance questions. Outcome engineers must treat agents as powerful delivery lanes that change ownership, review, and legal workflows (Principle 03, Principle 09).

Run OpenClaw Securely in Docker Sandboxes explains how to isolate always-on agents in Docker sandboxes and inject credentials via a proxy to prevent leaks. This is a concrete pattern for building safe local agent deployments and operationalizing the Gate and Tech Island ideas — sandboxing is non-negotiable for production agent workflows (Principle 07, Principle 10).

Ladybird adopts Rust, with help from AI reports a team using human-directed coding agents to port LibJS to Rust in two weeks with byte-for-byte identical outputs and zero regressions. That case shows agents can execute high-stakes migrations when combined with strong tests, human oversight, and conformance checks — a playbook outcome engineers can copy (Principle 03, Principle 14).