Architecture Driven AI Engineering
summary
Describes a five‑phase agentic software and IT lifecycle that enforces architecture-as-code via SysML v2, traces requirements into automated Gherkin acceptance tests, and runs agents with deterministic exit criteria plus human review gates. Reports small‑batch execution, versioned agents, skills, and workflows; instruments runtime with step timing, token/cost accounting and audit trails. Shows a seven‑person team replacing a $30M/year COTS tool and deploying full integrated ecosystems in minutes. Asks for a standardized AI package manager and deterministic agentic workflow standards.
AI-written summary · about
