Building Modern Distributed Systems using Less AI Tokens

Modern distributed systems are often hard to understand - not only for developers, but also for AI coding assistants. A simple business feature can quickly spread across frontend code, backend services, DTOs, REST or gRPC contracts, generated clients, SDKs, and integration layers. The result is more boilerplate, larger pull requests, higher token usage, and weaker architectural context for AI tools. In this hands-on workshop, participants will learn how to use skills-based AI workflows to build distributed systems that focus on business logic instead of integration code. We will show how a SKILLS.md file can describe reusable development capabilities, architectural constraints, and implementation patterns that AI assistants can follow consistently across the project. Participants will build a small modern distributed system with a frontend, backend, and service components while avoiding a traditional hand-written integration layer. Instead of generating large amounts of integration code, DTO mapping, API clients, and repetitive service contracts, the workflow will guide AI toward smaller, clearer changes that are easier to review and easier to understand. The workshop will demonstrate how to reduce lines of code, lower token usage, improve pull request readability, and make the system more understandable for both developers and AI assistants. By the end, participants will see how skills-driven development can help AI focus on what matters most: the business behaviour, system boundaries, and architectural intent.