
By the end of this lecture, students know:
what they will build;
what they will learn beyond the app;
whether the course is right for them;
why this is not a tool tutorial or a prompt collection.
Students understand:
the course has two connected tracks: the reusable system and the Pocket Budget case study;
the course is tool-independent;
each stage ends with evidence and a checkpoint;
Learn agentic engineering for Kotlin Multiplatform. In this course you build a real app with AI coding agents, using implementation contracts, test-first workflows, reviewer agents and evidence-based checkpoints. The app is Pocket Budget, a personal budgeting app that runs on Android, Desktop and iOS.
This is not a vibe coding course. You do not paste prompts and hope. You learn an engineering system that keeps AI coding agents inside clear limits, so the code they write is tested, reviewed and correct.
What you will build
Pocket Budget is built feature by feature, through a series of checkpoints:
Transactions saved with Room, with money stored as integer cents
Type-safe navigation and month browsing
Monthly budgets and category budgets
A financial dashboard
Authentication with Supabase and claiming local guest data
Dependency injection with Koin
An offline-first sync engine with an outbox, conflict resolution and Row Level Security
Runtime configuration for Desktop, Android and iOS
How you work with the AI agent
Turn each feature into an implementation contract with a clear scope and clear exclusions
Follow a test-first workflow: write a failing test, then the smallest code that passes
Build with focused skills for testing, navigation, design and offline sync
Use reviewer agents that check scope, architecture, domain rules, test evidence and remote-service risk
Run verification before every commit, and approve the result yourself
What you get
A starter project and a finished project to download
The full set of control files (contracts, skills, workflows and reviewers), added step by step as the course progresses
Git checkpoints, so you can compare your work at every stage
By the end, you can apply the same engineering system to another stack, project or AI coding environment.