Home Training
Jmix Course for Beginners

 Jmix for Beginners:
Build a Real App with an AI Agent

Go from an empty repository to a working Jmix application — not by clicking through wizards, but by writing specs, driving an AI agent, and understanding every line you ship.

Course Prerequisites

Basic Java
Relational databases (SQL, tables, relationships).
An IDE (IntelliJ IDEA recommended).
No prior Jmix experience required — we start from zero.

Built for Java Developers

build-for__card icon
Developers new to Jmix who want to build real applications with AI used the professional way, not "prompt and pray".
build-for__card icon
Engineers and tech leads evaluating Jmix who’d rather build a working app than watch a tour.
build-for__card icon
Anyone who tried "just ask the AI" and watched it drift, and wants a method that holds up.
basic-certificate image

After the course you will:

  • Build a complete Jmix application end to end -data model, UI, security, files, and a running, deployed app - and understand every part of it.
  • Drive a mature AI agent through a repeatable Spec → Agent → Review → Verify loop.
  • Tell where to trust the agent, where to use Jmix Studio, and where to fix by hand.
  • Read and correct generated code against how Jmix actually works - not just "does it compile".
At the end of the course, students receive a certificate of completion
and a badge that can be shared on social media. mascot image

Build a Real Application, Step by Step

One through-line project — an office meeting-room & desk booking app — built in
the order a real app comes together.

Every step runs the same loop:
1
Explain
the Jmix concept (what it is, why, how it works).
arrow right icon
2
Studio
generate one canonical example with Jmix Studio (the "golden sample").
arrow right icon
3
Spec
write a specification that embeds that example.
arrow right icon
4
Agent
the agent scales it to the rest.
arrow right icon
5
Review + Verify
check the result against how Jmix works, and run the app.
icon attention
You stay the author; the agent is an accelerator, not an autopilot.

Program (8 hours of training in English)

0
Setup & the spec-driven method
1
Data model & screens
2
Security
3
Working with files
4
Deployment
Module 0 - Setup & the spec-driven method
Course program:
  • Create a Jmix project and connect a mature AI coding agent to it (rules + skills).
  • The loop: Spec → Agent → Review → Verify.
  • Agent and Jmix Studio: where each is stronger (entity designer, view scaffolder, changelog from model diff) and how to choose.
Module 0 - Setup & the spec-driven method
Course program:
  • Create a Jmix project and connect a mature AI coding agent to it (rules + skills).
  • The loop: Spec → Agent → Review → Verify.
  • Agent and Jmix Studio: where each is stronger (entity designer, view scaffolder, changelog from model diff) and how to choose.
Module 1 - Data model & screens
Course program:
  • Domain modeling: entities, relationships, composition, enumerations.
  • Generating the model from a spec; Liquibase changelog; first launch.
  • List and detail screens, navigation, selecting references from a drop-down.
  • Business logic in a service; validating on save with a clear message.
  • Reuse what the framework already gives you (e.g. the built-in User) — catching over-modeling on review.
Module 2 - Security
Course program:
  • Authentication and authorization; the Jmix security subsystem.
  • Users, login, current-user information.
  • Resource roles, row-level roles, built-in roles, and composing roles.
Module 3 - Working with files
Course program:
  • Storing files: database vs file storage.
  • Upload field, download column in a table, manual file processing.
  • Multiple file storages and cloud storage (S3).
Module 4 - Deployment
Course program:
  • JAR and WAR formats, application name, properties.
  • Runtime profiles and profile-specific beans.
  • Docker and Docker Compose.
attention icon
Note: throughout, an honest agentic discipline — "compiles ≠ works", verify in the running app, and review the agent’s output for correctness, security, and maintainability.