<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>AI on SivaLabs</title><link>https://www.sivalabs.in/tags/ai/</link><description>Recent content in AI on SivaLabs</description><generator>Hugo</generator><language>en</language><lastBuildDate>Wed, 11 Nov 2026 12:31:08 +0530</lastBuildDate><atom:link href="https://www.sivalabs.in/tags/ai/rss.xml" rel="self" type="application/rss+xml"/><item><title>My AI Agentic Workflow</title><link>https://www.sivalabs.in/blog/my-ai-agentic-workflow/</link><pubDate>Tue, 10 Nov 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/my-ai-agentic-workflow/</guid><description>&lt;p&gt;I&amp;rsquo;d been getting good results with &lt;strong&gt;Opus 5.5&lt;/strong&gt; and &lt;strong&gt;GPT-6.1 Sol&lt;/strong&gt; in JetBrains Air, so I decided to take things a step further. I put a list of requirements in a Markdown file and used &lt;code&gt;/loop&lt;/code&gt; or &lt;code&gt;/goal&lt;/code&gt; to let the agents work through them.&lt;/p&gt;
&lt;p&gt;The agents finished implementing the requirements, but they made plenty of assumptions along the way. I ended up with a lot of code that only partially worked as intended. Untangling those assumptions and cleaning up the implementation looked like more work than rebuilding the use cases one at a time.&lt;/p&gt;
&lt;p&gt;So, my idea of &amp;ldquo;let&amp;rsquo;s build a software factory&amp;rdquo; is shelved for now :-)&lt;/p&gt;
&lt;p&gt;I see two main problems with handing AI a large batch of requirements:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;It&amp;rsquo;s easy to leave design decisions unstated.&lt;/strong&gt; Requirements that seem clear at first often hide questions about behavior, permissions, and edge cases. The agent fills in whatever we leave open.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;The resulting code is hard to review as a whole.&lt;/strong&gt; When so much changes at once, it&amp;rsquo;s difficult to build a clear understanding of how the system works and whether it behaves as intended.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;That brought me back to my tried-and-tested approach: &lt;strong&gt;one feature at a time&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Generating code is fast. Figuring out whether it builds what I wanted still takes work.&lt;/strong&gt;&lt;/p&gt;
&lt;p&gt;That&amp;rsquo;s the challenge my workflow tries to address. Here&amp;rsquo;s how I go from an idea to an implementation I can review without spending the whole day untangling assumptions.&lt;/p&gt;</description></item><item><title>Harness Engineering for Java and Spring Boot Applications</title><link>https://www.sivalabs.in/blog/harness-engineering-for-java-and-spring-boot-applications/</link><pubDate>Thu, 29 Oct 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/harness-engineering-for-java-and-spring-boot-applications/</guid><description>&lt;p&gt;I want to accelerate development using AI coding agents, but I don&amp;rsquo;t want to give up on code quality to get there. Generating code faster is useful only if I can keep the application understandable, maintainable, and correct as it grows.&lt;/p&gt;
&lt;p&gt;An agent can implement a feature, generate tests, and refactor several classes in a few minutes. Reviewing all those changes carefully still takes time. If every review requires me to point out formatting issues, explain package boundaries again, and ask for missing authorization tests, I have moved the bottleneck rather than removed it.&lt;/p&gt;</description></item><item><title>Spring AI : Advisors API</title><link>https://www.sivalabs.in/blog/spring-ai-advisors-api/</link><pubDate>Thu, 30 Jul 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/spring-ai-advisors-api/</guid><description>&lt;p&gt;Once an AI feature moves beyond a demo, the model call is rarely the whole story.&lt;/p&gt;
&lt;p&gt;You need to log requests, restore conversation history, retrieve relevant documents, reject unsafe input, record metrics, and perhaps validate the model&amp;rsquo;s output.
Putting all of that code directly around every &lt;code&gt;ChatClient&lt;/code&gt; call works for a while. Then a second endpoint arrives, followed by streaming, and suddenly every model call has its own slightly different pile of plumbing.&lt;/p&gt;</description></item><item><title>Spring AI : Chat Memory</title><link>https://www.sivalabs.in/blog/spring-ai-chat-memory/</link><pubDate>Sat, 25 Jul 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/spring-ai-chat-memory/</guid><description>&lt;p&gt;Large language models are stateless. Send two independent requests to a model, and the second request knows nothing about the first one.
That is perfectly fine for one-shot questions, but it makes a chatbot feel surprisingly forgetful:&lt;/p&gt;</description></item><item><title>Spring AI : Structured Output</title><link>https://www.sivalabs.in/blog/spring-ai-structured-output/</link><pubDate>Wed, 22 Jul 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/spring-ai-structured-output/</guid><description>&lt;p&gt;Chat models are great at producing prose, but most applications don&amp;rsquo;t want prose — they want a Java object, a &lt;code&gt;List&amp;lt;String&amp;gt;&lt;/code&gt;, or a &lt;code&gt;Map&lt;/code&gt; they can bind to a response and return from a REST API.&lt;/p&gt;
&lt;p&gt;Spring AI&amp;rsquo;s &lt;strong&gt;Structured Output Converters&lt;/strong&gt; bridge that gap: they turn plain-text chat responses into typed Java objects, without you having to hand-write a JSON parser or repeat &amp;ldquo;please respond with valid JSON&amp;rdquo; in every prompt.&lt;/p&gt;</description></item><item><title>Spring AI : Chat with LLMs</title><link>https://www.sivalabs.in/blog/spring-ai-chat-with-llms/</link><pubDate>Mon, 20 Jul 2026 06:00:00 +0530</pubDate><guid>https://www.sivalabs.in/blog/spring-ai-chat-with-llms/</guid><description>&lt;p&gt;Calling an LLM is just an HTTP request. The trouble is that every provider has its own API,
configuration, and response format.&lt;/p&gt;
&lt;p&gt;Spring AI handles that provider-specific plumbing and gives us a familiar Spring API. In this article,
we will use &lt;code&gt;ChatClient&lt;/code&gt; to talk to OpenAI&amp;rsquo;s GPT-5 model and then make our prompts reusable with
&lt;code&gt;PromptTemplate&lt;/code&gt;.&lt;/p&gt;</description></item><item><title>AI Agents Are Doing The Heavy Lifting, Then Why Am I So Tired?</title><link>https://www.sivalabs.in/blog/ai-agents-are-doing-the-heavy-lifting/</link><pubDate>Tue, 30 Jun 2026 04:59:17 +0530</pubDate><guid>https://www.sivalabs.in/blog/ai-agents-are-doing-the-heavy-lifting/</guid><description>&lt;p&gt;I love writing code.&lt;/p&gt;
&lt;p&gt;Not just because it pays the bills.&lt;/p&gt;
&lt;p&gt;I genuinely enjoy the process.&lt;/p&gt;
&lt;p&gt;Give me a use case, let me think about the design, write a failing test, make it pass, refactor, move classes around,
rename things, delete half the code I wrote an hour ago, run the test suite again, smile because everything is still green&amp;hellip;
and I can happily spend an entire day doing that.&lt;/p&gt;</description></item><item><title>Confessions of an AI Agent FOMO Survivor</title><link>https://www.sivalabs.in/blog/confessions-of-an-ai-agent-fomo-survivor/</link><pubDate>Mon, 22 Jun 2026 04:59:17 +0530</pubDate><guid>https://www.sivalabs.in/blog/confessions-of-an-ai-agent-fomo-survivor/</guid><description>&lt;p&gt;Every morning, I open X/Twitter, LinkedIn, or YouTube and learn about a brand-new AI engineering discipline that apparently everyone else has already mastered.&lt;/p&gt;
&lt;p&gt;A few months ago it was Prompt Engineering.&lt;/p&gt;
&lt;p&gt;Then came Context Engineering.&lt;/p&gt;
&lt;p&gt;Soon after, people started talking about &lt;strong&gt;Agent Skills, Harness Engineering, Loop Engineering, Ralph Wiggum Loops, Sub-Agent Architectures, Agentic Workflows, Cognitive Tooling&lt;/strong&gt;, and probably three more terms while I was writing this sentence.&lt;/p&gt;
&lt;p&gt;At this point I fully expect to wake up tomorrow and discover that &amp;ldquo;Recursive Quantum Harness Context Loop Engineering&amp;rdquo; is the new must-have skill for AI developers.&lt;/p&gt;</description></item><item><title>Before You Judge, Get Curious</title><link>https://www.sivalabs.in/blog/before-you-judge-get-curious/</link><pubDate>Wed, 22 Apr 2026 04:59:17 +0530</pubDate><guid>https://www.sivalabs.in/blog/before-you-judge-get-curious/</guid><description>&lt;p&gt;A few years ago, I was working at a fintech company in Germany. Our business unit planned to build a new offering for customers in Germany and the UK.
Our team — product owner, business analyst, developers, and QA — gathered at a location and ran brainstorming sessions.
Since we were also using the product ourselves, we were essentially both the builders and the end users.&lt;/p&gt;</description></item><item><title>Feeling Left Behind in Tech? This Is Your 90-Day Comeback Plan</title><link>https://www.sivalabs.in/blog/feel-left-behind-in-tech-this-is-90-day-comeback-plan/</link><pubDate>Wed, 18 Mar 2026 04:59:17 +0530</pubDate><guid>https://www.sivalabs.in/blog/feel-left-behind-in-tech-this-is-90-day-comeback-plan/</guid><description>&lt;p&gt;You&amp;rsquo;ve been working for years. You know your systems inside out.
You&amp;rsquo;ve solved real problems, handled pressure, and delivered results.&lt;/p&gt;
&lt;p&gt;But now you open a job description… and it feels like you&amp;rsquo;re reading a different language.&lt;/p&gt;
&lt;p&gt;New frameworks. New tools. New expectations.&lt;/p&gt;
&lt;p&gt;And the thought creeps in: &lt;em&gt;&amp;ldquo;Have I fallen behind?&amp;rdquo;&lt;/em&gt;&lt;/p&gt;</description></item><item><title>Stepping into Crazy AI World</title><link>https://www.sivalabs.in/blog/stepping-into-crazy-ai-world/</link><pubDate>Thu, 22 Jan 2026 04:59:17 +0530</pubDate><guid>https://www.sivalabs.in/blog/stepping-into-crazy-ai-world/</guid><description>&lt;p&gt;If you are feeling overwhelmed by the rapid pace of technological advancements, especially in the field of AI, know that you are not alone.
The AI landscape is evolving at an unbelievable speed, with new ideas and tools emerging every day.&lt;/p&gt;</description></item><item><title>You Can Code Offline With Local AI(Ollama)</title><link>https://www.sivalabs.in/blog/code-offline-with-local-ai-ollama/</link><pubDate>Thu, 09 Jan 2025 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/code-offline-with-local-ai-ollama/</guid><description>&lt;p&gt;ChatGPT and other Generative AI tools took the world by a storm.
People are using these AI tools for various purposes such as to explore a topic,
to seek answers to their questions or to get help in coding, etc.
ChatGPT and Google&amp;rsquo;s Gemini are very popular AI tools that are available for free with some usage limitations.&lt;/p&gt;</description></item><item><title>My First Impression of Google's NotebookLM</title><link>https://www.sivalabs.in/blog/my-first-impression-of-notebooklm/</link><pubDate>Fri, 15 Nov 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/my-first-impression-of-notebooklm/</guid><description>&lt;p&gt;I came across &lt;a href="https://notebooklm.google/"&gt;NotebookLM&lt;/a&gt; from my social media feed, and it looks interesting.&lt;/p&gt;
&lt;p&gt;With NotebookLM, you can upload documents, videos, web URLs and let the AI agents talk about the uploaded material.
That sounds intriguing.&lt;/p&gt;
&lt;p&gt;Recently I wrote an article about software complexity and shared my viewpoint.
Many people appreciated it and some people didn&amp;rsquo;t. That is fine to have different opinions.&lt;/p&gt;</description></item><item><title>Spring AI RAG using Embedding Models and Vector Databases</title><link>https://www.sivalabs.in/blog/spring-ai-rag-using-embedding-models-vector-databases/</link><pubDate>Thu, 14 Mar 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/spring-ai-rag-using-embedding-models-vector-databases/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Introduction to Embedding Models.&lt;/li&gt;
&lt;li&gt;Loading data using DocumentReaders.&lt;/li&gt;
&lt;li&gt;Storing embeddings in VectorStores.&lt;/li&gt;
&lt;li&gt;Implementing RAG (Retrieval-Augmented Generation), a.k.a. Prompt Stuffing.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item><item><title>Getting Started with Spring AI and OpenAI</title><link>https://www.sivalabs.in/blog/getting-started-with-spring-ai-and-open-ai/</link><pubDate>Sat, 09 Mar 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/getting-started-with-spring-ai-and-open-ai/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Introduction to Spring AI.&lt;/li&gt;
&lt;li&gt;Interacting with OpenAI using Spring AI.&lt;/li&gt;
&lt;li&gt;Using &lt;strong&gt;PromptTemplates&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Using &lt;strong&gt;OutputConverters&lt;/strong&gt;.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item><item><title>LangChain4j Retrieval-Augmented Generation (RAG) Tutorial</title><link>https://www.sivalabs.in/blog/langchain4j-retrieval-augmented-generation-tutorial/</link><pubDate>Sun, 25 Feb 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/langchain4j-retrieval-augmented-generation-tutorial/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Understand the need for &lt;strong&gt;Retrieval-Augmented Generation (RAG)&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Understand &lt;strong&gt;EmbeddingModel&lt;/strong&gt;, &lt;strong&gt;EmbeddingStore&lt;/strong&gt;, &lt;strong&gt;DocumentLoaders&lt;/strong&gt;, &lt;strong&gt;EmbeddingStoreIngestor&lt;/strong&gt;.&lt;/li&gt;
&lt;li&gt;Working with different EmbeddingModels and EmbeddingStores.&lt;/li&gt;
&lt;li&gt;Ingesting data into EmbeddingStore.&lt;/li&gt;
&lt;li&gt;Querying LLMs with data from EmbeddingStore.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item><item><title>LangChain4j AiServices Tutorial</title><link>https://www.sivalabs.in/blog/langchain4j-ai-services-tutorial/</link><pubDate>Thu, 22 Feb 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/langchain4j-ai-services-tutorial/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Using LangChain4j &lt;strong&gt;AiServices&lt;/strong&gt; to interact with LLMs.&lt;/li&gt;
&lt;li&gt;How to ask questions and map responses to different formats?&lt;/li&gt;
&lt;li&gt;Summarizing the given text in different formats.&lt;/li&gt;
&lt;li&gt;Analyzing the sentiment of the given text.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item><item><title>Generative AI Conversations using LangChain4j ChatMemory</title><link>https://www.sivalabs.in/blog/generative-ai-conversations-using-langchain4j-chat-memory/</link><pubDate>Tue, 20 Feb 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/generative-ai-conversations-using-langchain4j-chat-memory/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;How to use LangChain4j &lt;strong&gt;ChatMemory&lt;/strong&gt; and &lt;strong&gt;ConversationalChain&lt;/strong&gt; to implement conversation style interaction?&lt;/li&gt;
&lt;li&gt;How to ask questions using &lt;strong&gt;PromptTemplate&lt;/strong&gt;?&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;In the &lt;a href="https://www.sivalabs.in/blog/getting-started-with-generative-ai-using-java-langchain4j-openai-ollama/"&gt;previous article&lt;/a&gt;, we have seen how to interact with OpenAI using Java and LangChain4j.&lt;/p&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item><item><title>Getting Started with Generative AI using Java, LangChain4j, OpenAI and Ollama</title><link>https://www.sivalabs.in/blog/getting-started-with-generative-ai-using-java-langchain4j-openai-ollama/</link><pubDate>Sun, 18 Feb 2024 23:29:17 +0000</pubDate><guid>https://www.sivalabs.in/blog/getting-started-with-generative-ai-using-java-langchain4j-openai-ollama/</guid><description>&lt;p&gt;In this article, we will explore the following:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;Brief introduction to Generative AI?&lt;/li&gt;
&lt;li&gt;How to interact with OpenAI APIs using Java?&lt;/li&gt;
&lt;li&gt;How to use LangChain4j to interact with OpenAI?&lt;/li&gt;
&lt;li&gt;How to run a LLM model locally using Ollama?&lt;/li&gt;
&lt;li&gt;Working with Ollama using LangChain4j and Testcontainers.&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Let&amp;rsquo;s get started.&lt;/p&gt;</description></item></channel></rss>