The glossary
The vocabulary behind how we build. See these ideas in context on our approach page.
- IT 2.0
- Scutiger’s name for a modern software services model that uses agentic AI to ship production software in weeks, replacing sprint theater, PRD handoffs, and manual QA gates.
- Agentic Delivery
- A way of building software in which AI agents handle scaffolding and repetitive code while engineers focus on architecture, verification, and hard problems.
- Two-Lane Model
- A delivery structure with a fast prototype lane (AI-generated) running alongside a hardened production lane (verified), so speed never compromises reliability.
- Prototype Lane
- The fast lane of the two-lane model, where AI-generated prototypes validate ideas and flows in days before any commitment to a full build.
- Production Lane
- The hardened lane of the two-lane model, where verified software runs in production backed by continuous guardrails.
- Hardening Gate
- An explicit transition point where a prototype is evaluated for production readiness — security, reliability, performance, and compliance — before it goes live.
- Guardrails Over Gates
- A governance principle that replaces manual end-of-line QA gates with continuous automated quality, security, and compliance checks.
- Golden Paths
- Pre-approved, paved development routes — vetted tools, templates, and pipelines — that make compliance and best practice the default, automatic choice.
- No-Handoff Methodology
- A way of working in which product, design, and engineering co-create from day one instead of passing documents over the wall between teams.
- Executable Intent
- The idea that a working prototype, not a comprehensive document, is the clearest shared expression of what a team intends to build.
- Sprint Theater
- The ceremony-heavy, story-point-obsessed rituals of traditional agile that produce incremental outputs without necessarily producing outcomes — the practice IT 2.0 replaces.
- AI Agent
- An AI system that can carry out multi-step software tasks — generating scaffolding, boilerplate, and repetitive code — under the direction of an engineer.
- MLOps
- The practice of taking machine-learning models from notebook to production reliably, using pipelines, feature stores, and model-serving infrastructure.
- Platform Engineering
- Building internal developer platforms — golden paths, design systems, and shared infrastructure — that let teams scale quality across many projects.
- Feature Store
- A system that manages and serves the data features used to train and run machine-learning models consistently across development and production.
- Guardrail
- An automated check — for quality, security, or compliance — that runs continuously during development to catch issues as they appear rather than at the end.
Want to see these in practice?
Read our approach, browse the handbooks, or tell us about your project.