Spring AI Introduction: Building AI Applications in Java with Spring

For the longest time if you wanted to work with Machine Learning you were using Python. That was because python had a lot a really great set of tools for working in this space. Now that we have LLMs like Open AI’s GPT-4 where the P stands for pre-trained we can simply call them like we would any other API. Java is everywhere in the enterprise and we need to start thinking about how to integrate these capabilities into our applications.
In today’s tutorial I want to talk about communicating with an LLM like GPT-4 (or any other LLM) and some of the challenges you might come across. From there I’ll introduce you Spring AI which is going to help us build AI powered applications by solving some of these challenges.
🔗Resources & Links mentioned in this video:
👋🏻Connect with me:
Website: www.danvega.dev
Twitter: / therealdanvega
Github: github.com/danvega
LinkedIn: / danvega
Newsletter: www.danvega.dev/newsletter
SUBSCRIBE TO MY CHANNEL: bit.ly/2re4GH0 ❤️

[@fibfranco]

Such an amazing coincidence haha
I started looking at the Spring AI project for the first time a couple hours ago and thought: "wonder if vega has posted something about it" and you just did :)
thank you, man!

[@DanVega]

That is funny 😄 I certainly don’t have all of my subscribers phones tapped 😂

[@sohamsarkar174]

Dan loved your session. Please upload a detailed video on RAG

[@theunthinkable6280]

Top, Clear, Informative. Good job!

[@TobilobaOwolabi-wy8cq]

Hi Dan. Thank you for the subtle intro to Spring AI. I know you will be looking to do more videos on this topic and I would like to just point out some topics you can look into.
1. RAG with multiple documents instead of one
2. How to use more than one LLM and switch between them(maybe with dynamic configuration properties)

[@MohamedDernoun]

How clear is your explanation, best address for people who scare from java!

[@user-dn8nm9ie2y]

Awesome @Dan Vega!! Keeping it simple but effective. Gave me the needed high level overview that I needed

[@younaspa1058]

Thanks for the wonderful session

[@CodeBank-rw9bq]

Exactly this is what I was looking for ❤

[@WisomofHal]

Oh my goodness. In my current Springboot project I've created a complete "Python Engine", using Java's process builder, to create embedding, parse embeddings, run all the required python scripts, error handling and output to create and run RAG Chains. It was quite fun, but this will help tremendously to take that load off. Thanks for sharing will be looking deeper into this. Cheers!

[@douglasdotv]

Thank you for another video.

[@minarashidi868]

Thanks, that was so inspiring!

[@backtoGodhead0]

so cool, God bless you!

[@stephaneislistening6103]

Very cool Thanks !

[@scitechplusexplorer2484]

Hey Dan, hello, Good Evening! Please make a full video on GraphQL including Pagination with latest Spring Boot version. It would be very helpful.

[@DanVega]

I have done a video on pagination you can watch it here
ruclips.net/video/3YTSh8vJ8eY/видео.html

[@walterpalladino1965]

Great video. Can Spring AI interact with a local LLM? If can do it, will be great to se an example. Thanks.

[@sagarbhat1]

Thanks for taking time to make this video Dan.
I’m building a RAG based application and have tried LangChain4j and Spring AI, and at this stage LangChain4j seems to have more features (Chat Memory, Hybrid semantic search with Azure AI search vector store) compared to Spring AI. But at the rate at which Spring AI is being developed, I’m sure it will catch up.
Quick question - is there an easy way to log the requests and responses to/from the Open AI service in Spring AI? In Langchain4j it’s as simple as adding a property.

[@DanVega]

Spring AI in an MVC stack is using the RestClient. You need to enable logging for what ever underlying http client you're using which by default is the HttpUrlConnection. You can set this logging level to see the request/response. I would take this one step further though and add the actuator where you can then get metrics on all of these calls.
logging.level.sun.net.www.protocol.http.HttpURLConnection=DEBUG

[@sagarbhat3884]

@@DanVega I tried this (logging.level.sun.net.www.protocol.http.HttpURLConnection=DEBUG) but it did not help. I could not see any trace of the request and response to/from the AI model.

[@gsilveralan]

Geat video please do one with RAG

[@Bibhaw]

Well explained, Thanks vega :)
Can't we directly autowired ChatClient ?

[@terrabys]

Is there support for agentic workflow?

[@BananBananski]

Looks like the ChatClient is synchronous and some models take a long time to respond - that would be a bottleneck in many use cases. Is there an asynchronous implementation as well?

[@DanVega]

Yes there is a StreamingChatClient
docs.spring.io/spring-ai/reference/api/chatclient.html#_streamingchatclient

[@muslehhaj4164]

Hello Spring Boss 💪, I have a course idea but no one in "Udemy" did it yet K I've tried few course in Udemy , but none of them completed... I'm looking for a full-stack project wise course using spring boot as backend & Vaadin for front end .... but not sure if Vaadin is the right choice to build a real-life application .... please Advise me Boss ⚠

[@adityasinghrajput7815]

Sir please upload a full spring boot beginner to advanced series and after that spring security then spring cloud then micro servise then full project with spring boot integration of spring AI

[@checkantetibo]

Interesting

[@crampanelarutzi]

please make a video about weka framework.

[@kafkaesqued]

Hello Sir, could you provide a reference for SVN to GitHub migration with the help of Spring Java + GenAI tech

[@stephaneislistening6103]

On SpringBoot 3.3.0 here and the OpenAI API seems to have changed quite a bit already in a month since this video was published.

[@stephaneislistening6103]

And so the chatClient bean could not be found, with the default implementation being: class DefaultChatClient implements ChatClient

[@davidwaziri333]

Can you use free ChatGPT API key for this?

[@Gijox1]

Very nice!

[@antonslonkin]

Interesting!!!