Codepocalypse Now: LangChain4j vs. Spring AI
A presentation at DevoxxFR in in Paris, France by Viktor Gamov
Codepocalypse Now LangChain4j vs. Spring AI
@jbaruch http://speaking.jbaru.ch @gamussa https://speaking.gamov.io/
Why Frameworks? / Devoxx France Codepocalypse Now LangChain4j vs. Spring AI
Thank You, DEVOXX FRANCE
Prompt-Driven Development: Aligning Ideas, Tests, and Code @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
We have a Trust problem @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
AI generated code is not great @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
On top of that, it is dangerou s @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Asking it to fix it is as reliabl e as the rest of it @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Baruch Sadogursky - @jbaru ch × Developer Advocate at large (talk to me!) × Development -> DevOps -> #DPE @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
shownotes × × × × speaking.jbaru.ch Slides Video All the links! @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Software design documents @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Software design documents × Write-once × Read-maybe-once × (Mis)understood by humans @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
So why don’t they work? × Human (mis)understanding × Vague responsibility boundaries @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
We have a Trust problem (not only with AI) @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Next thing you know: It’s a vein Software I know really well Software I like @jbaruch @gamussa diagram @Devoxxfr #pdd speaking.jbaru.ch
But hey, we do have working s oftware sometimes × Good intentions × Professionalism × Tests and QA × End result observation @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Until gen AI changed the game × × × × Good intentions Professionalism Tests and QA End result observation @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
AI Coding Agents Software quality @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
I find your lack of trust disturbing @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Idea! Verifiable intent What if we code in the intent and always verify against it? @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Let’s express intent in tests × × × × Instead of SDDs Always up-to-date Generate consensus Parsed by the machine @jbaruch @gamussa @Devoxxfr TDD? #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
What if it (almost) won’t look × × × × × like code? Instead of SDDs Always up-to-date Generate consensus Parsed by the machine Describe behavior instead of tests @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
What if it (almost) won’t look × × × × × like code? Always up-to-date BDD? Generate consensus Parsed by the machine Describe behavior instead of tests Almost plain English human language @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
@jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
But writing tests first is still Business people won’t write tests @jbaruch a good idea, right?! But the developers Still will, right?! @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Why developers won’t write te sts first × Developers are solution-biased × We already know how to solve the problem × Legacy code bases @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
LLMs to the rescue! @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
How everything comes togeth er Business people define what software should do @jbaruch LLM creates specs @gamussa Specs are reviewed and agreed upon @Devoxxfr Tests are generate d and reviewed #pdd Code is written to satisfy the tests -> specs -> design speaking.jbaru.ch
Why does it work? × × × × × Business people write text Specs are reviewed by all SDDs are Specs are living docs Everything else is derived from specs Previous steps are protected @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
PDD is BDD as it meant to be × Define and agree on intent × AI is protected from circular verification × Up-to-date context docs at all times @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Rules/Guidelines × Mostly refined by Cascade × The full version covers an existing codebase scenario × Has a link in /docs @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
New project × Phase 1: Define requirements using the memento pattern × Phase 2: Define test scenarios and specs × Phase 3: Implement tests and business logic @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Existing project × Phase 0: Examine code and produce full test coverage × Existing tests are protected @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Protection rules × Requirement documents are protected after Phase 1. × If changes are needed, the phase is reset × Test scenarios and specs are protected after Phase 2. @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Recovery rules × Source control commits after stages × Requirements changes trigger phase reset and re-alignment × Might convert the project into ”existing” mode @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Cascade in Windsurf Pros × Speed × Conversation and reflection × Selectable models @jbaruch @gamussa Cons × Not IntelliJ IDEA × .windsurfrules system is not ideal @Devoxxfr #pdd speaking.jbaru.ch
Junie in intellij idea Pros × IntelliJ IDEA × Execution Plan × guidelines.md is better @jbaruch @gamussa Cons × Speed × Introversion × No chat/reflection mode @Devoxxfr #pdd speaking.jbaru.ch
Better Rules system? × × × × × Multiple files for different scenarios Flexible actuators File references for reuse Optionally, part of the documentation .cursor/rules for the best implementation ATM @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Too much code? × New prompt – new context window × Memento pattern FTW × But what if even the initial context is too large? × Is it the perfect argument for microservices? @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Who should fix the code? × Code inspections (linting in VS Code) integration × “Switch from AI generation to tool calling when needed” × Leonid Kuligin – “What is an AI agent? How to develop one?” @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Developer productivity @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Is it engaging? @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Looks like a pretty easy fix… @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Are the stops intentional? ”rich mode”? @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Use tools, add guardrails Using IntelliJ IDEA code inspections @jbaruch @gamussa Machineverify stages compliance Physically protect stage assets @Devoxxfr #pdd speaking.jbaru.ch
Is it too rigid? × The process feels the opposite of agile × Making changes is a pain × Are there other options? @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
(Summary) PDD – BDD finally m a kes sense × Generates consensus × Features are verifiable back to requirements × We can start trusting AI code (🤞) × But there is still work to be done @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
THANKS! Q&A and Twitter X/Bsky/LinkedIn ads: x x x x x @jbaruch @gamussa @devoxxfr #pdd speaking.jbaru.ch @jbaruch @gamussa @Devoxxfr #pdd speaking.jbaru.ch
Which Java framework handles AI better: LangChain4j or Spring AI? In this live coding showdown, we’ll build a semantic code search application from scratch, putting both frameworks to the test. We’ll cover project setup, using language models, setting up a retrieval-augmented generation workflow, and creating a REST API.
You’ll see how these frameworks handle embedding generation, vector database integration, and real-world development challenges. By the end, the audience decides who wins—based on which framework gets the job done faster, better, and with less hassle (and whether the demo actually works).
This isn’t just a comparison; it’s a live experiment under pressure. Come and see which one comes out on top!
Video
Resources
The following resources were mentioned during the presentation or are useful additional information.
Code
The following code examples from the presentation can be tried out live.
Buzz and feedback
Here’s what was said about this presentation on social media.
