标签：#AI

Most SaaS backends were built around a simple assumption: The user pays a subscription, then uses the product. That assumption breaks down for AI apps. An AI app does not just serve screens. It spends money while it works. A user searches the web. A model summarizes a report. An image model generates a draft. An agent calls an MCP tool. A workflow buys data from an API. A future x402 endpoint charges for a capability call. Every one of those actions can have a marginal cost. That means the backend for an AI app is no longer just a place to store users, projects, and settings. Increasingly, it is a system of record for economic activity. In other words: AI app backends are becoming accounting systems. The old SaaS model was simpler Traditional SaaS could survive with coarse billing. You had: monthly subscriptions seats tiers maybe a usage limit somewhere That worked because the marginal cost of most product actions was close enough to zero. If a user clicked a button, edited a document, opened a dashboard, or created a project, the backend cost was usually small compared with the subscription price. The business could average it out. AI apps are different. The product may call paid APIs on almost every useful action. Search once. Summarize once. Generate once. Transcribe once. Call an agent tool once. The unit economics are inside the interaction loop. If the product cannot see who spent what, when, and why, the business is flying blind. Usage billing is not an add-on For AI apps, usage billing is often treated like a pricing feature. I think that is too narrow. Usage billing is really a cost ledger. It answers: which user triggered the cost? which project or app did it belong to? which capability was called? what did it quote before execution? what did it actually cost? was it retried? was it idempotent? did the end user pay for it? is there a payment or checkout record attached? If you cannot answer those questions, you do not have a reliable production backend for

We’ve all been there: it’s 7:00 PM, you’re exhausted after a long sprint, and you open a food delivery app. Your brain screams "Double Cheeseburger," but your body is still recovering from that mid-afternoon sugar spike. What if your phone was smart enough to say, "Hey, your blood sugar is currently 160 mg/dL and rising—maybe skip the extra fries?" In this tutorial, we are building a Chief Health Officer (CHO) Agent . This isn't just a simple chatbot; it’s a sophisticated AI Agent using LangGraph to bridge the gap between real-time medical data (CGM) and real-world actions (Food Delivery APIs). By leveraging automation , function calling , and state machines , we’ll create a system that actively protects your metabolic health. The Architecture: How the CHO Agent Thinks To build a reliable agent, we need a "stateful" workflow. We aren't just sending a prompt to an LLM; we are creating a loop that monitors glucose levels, analyzes food options, and interacts with the browser. graph TD A[Start: Hunger Trigger] --> B{Fetch CGM Data} B -->|Sugar High/Unstable| C[Constraint: Low GI Only] B -->|Sugar Stable| D[Constraint: Balanced Meal] C --> E[Scrape Delivery App Menu] D --> E E --> F[Agent: Analyze Ingredients & GI Index] F --> G[Selenium: Mark/Filter Non-Compliant Items] G --> H[End: Safe Ordering] subgraph "The LangGraph Loop" C D E F end Prerequisites Before we dive into the code, ensure you have the following: LangGraph & LangChain : For the agent's cognitive architecture. Dexcom API Credentials : To fetch real-time Continuous Glucose Monitor (CGM) data. Selenium : For interacting with food delivery web interfaces (Meituan/Ele.me). OpenAI API Key : Specifically for GPT-4o’s reasoning and function-calling capabilities. Step 1: Defining the Agent State In LangGraph, everything revolves around the State . Our CHO agent needs to track the current glucose level, the user's health constraints, and the list of available food items. from typing import TypedDict , List , Anno

With over 5 years in customer support and retention, I've lost count of how many times I've seen the same pattern: a customer explains an issue, gets it "resolved," and then has to explain the same problem again weeks later, as if the first conversation never happened. Support systems forget. Customers don't. That frustration, seen over years on the support side, is what led me to this hackathon project. Most support systems and most AI chatbots treat every interaction as isolated. They don't remember. So patterns that should be obvious (repeated complaints, dropping usage, unresolved issues) never get connected until a customer just leaves. That became the seed for my project: the Retention Risk Agent. The Problem With "Forgetful" AI Most AI tools answer questions in the moment, then forget everything. Ask a chatbot about a customer's history, and it only knows what's in that single message, not what happened last week, last month, or across five different support tickets. For churn prediction, that's a fatal flaw. Churn isn't a single event. It's a pattern, a series of small signals that only make sense when viewed together over time. This is something I understand deeply from years of watching it happen firsthand. Cognee is an open-source memory layer for AI agents. Instead of treating each interaction as isolated, it builds a knowledge graph, connecting facts, relationships, and context across everything you feed it. That's exactly what churn detection needed. What I Built I created a Python script that: Ingests customer records (support tickets, usage patterns, plan changes) Uses Cognee to build a memory graph connecting these signals Asks a simple question: "Which customers show signs of churn risk, and why?" The result wasn't a keyword match; it was reasoning. The agent correctly flagged a customer whose usage dropped 80% and who'd ignored two check-in emails. It flagged another who'd complained twice about slow support and mentioned a competitor. And critica

I built a 36-pattern checklist to catch AI writing tells in my own drafts, calibrated against...

Most Americans don't trust AI. It's proven that it doesn't know what safe toppings for pizza are. People don't even want to listen to AI music. But none of that matters for some of America's wealthy, who are turning to AI to teach their kids instead of traditional schools. Companies like Forge Prep and Alpha […]

Here's the failure mode I keep running into. A team gives a coding agent a repo, a task, and maybe a README. The agent can find files and write code, but it still has to guess the operating rules. It guesses the package manager. It guesses which checks matter. It guesses whether generated files are safe to edit. It guesses what "done" means. A README is usually for humans: what the project is, how to run it, and where the important docs live. A coding agent needs different context. Setup rules. Test commands. Boundaries. Completion criteria. That's the gap AGENTS.md fills. The official AGENTS.md guidance describes it as a predictable place for coding-agent instructions: setup commands, test commands, code style, security considerations, and nested instructions for large monorepos. I find the split useful in a more boring way. The README answers, "What is this project?" AGENTS.md answers, "What should an agent know before touching it?" That second question is where the work usually gets fragile. Where Goose Fits Goose makes this less theoretical because it isn't just a chat box. It's an open source local AI agent with a desktop app, CLI, API, MCP extensions, and skills. Without AGENTS.md , I find myself writing prompts like this: Update the docs, but don't touch generated files, use pnpm, run the lint and test commands, keep the PR small, and tell me what you couldn't verify. With AGENTS.md , the prompt can get shorter: Update the quickstart docs for the new config flag. Goose can run the task in the repo. The repo can carry the standing instructions. I noticed this on a small docs/config update where generated files sat near source files. Without repo instructions, the prompt had to carry the package manager, generated-file boundary, checks, and the "tell me what you could not verify" rule. Once those rules lived in AGENTS.md , the prompt became just the task. Not magic. Just fewer chances to forget the boring parts. Where Skills Fit I would add one more layer once

# Sakana Fugu: The Multi-Agent AI System That Works Like a Team We’ve all been there: copy-pasting a prompt from ChatGPT to Claude, and then to Gemini, trying to find which AI gives the best answer. Different AI models have different strengths. Some are excellent programmers, some write beautifully, and others are master logicians. But constantly switching between them is slow and frustrating. Sakana AI has introduced a brilliant solution: Sakana Fugu . Fugu is a multi-agent AI system that bundles multiple frontier models into a single, seamless package. To you, it looks like a single chatbot. But behind the scenes, it acts as a project manager, coordinating a team of top-tier AI models to solve your task. What is Sakana Fugu? Sakana Fugu is a collaborative AI framework. Instead of relying on one massive AI model to do all the work, Fugu orchestrates a pool of specialized models (such as GPT-5, Claude Opus, and Gemini Pro). When you give Fugu a complex, multi-step prompt, it: Analyzes the task and breaks it down into steps. Assigns specialized roles (like researcher, coder, and editor) to different AI models. Passes the work back and forth between them until the final, polished result is ready. By working as a team, these models achieve far better results than any single AI could on its own. The Secret Sauce: Teamwork Over Size Fugu’s intelligent coordination is based on two core concepts presented at the ICLR 2026 conference: 1. The Manager (TRINITY) Think of TRINITY as the manager of the team. It is a compact coordinator model that assigns specific roles to the worker LLMs. For example, it might tell Gemini to generate the initial code, tell Claude to find the bugs, and tell GPT to write the user documentation. They collaborate seamlessly without needing to merge their code bases. 2. The Conductor The Conductor acts as the communication network designer. It figures out the best way for the worker AIs to talk to each other for your specific question. It writes cust

Introduction Although we interact with LLMs using natural language, these models never processes raw text directly. Before a prompt reaches the model, it is converted into a sequence of tokens , the fundamental units that the model understands. Tokenization is one of the earliest stages of the inference pipeline and influences everything from context windows and API pricing to latency and memory usage. What Is a Token? A token is the smallest unit of text processed by a language model, it is not necessarily a word. Depending on the tokenizer, a token may represent: an entire word part of a word punctuation whitespace numbers symbols emojis Different models use different tokenizers, so the same text may be split differently depending on the model. Why Tokens? Simply because language models operate on numbers, not text. Before the transformer can perform any computation, the input must be converted into a numerical representation. The preprocessing pipeline looks like this: Raw Text │ ▼ Tokenizer │ ▼ Tokens │ ▼ Token IDs │ ▼ Embedding Layer │ ▼ Embedding Vectors │ ▼ Transformer The tokenizer splits the input into tokens and each token is then mapped to a unique integer called a token ID , which are passed through the model's embedding layer, which converts them into dense vectors that become the actual input to the transformer. A Real Example Instead of using hypothetical examples, let's look at how OpenAI's tokenizer processes text. Input: I have no enemies. OpenAI tokenizes it to: ["I", " have", " no", " enemies", "."] with the following token IDs: [40, 679, 860, 33974, 13] that have been generated by OpenAI Tokenizer for the "GPT-5.x & O1/3" models. The transformer never sees the original sentence, it only receives the corresponding sequence of token IDs. Token IDs After tokenization, every token is replaced with an integer. Conceptually: " have" → 679 " no" → 860 " enemies" → 33974 ... The exact numbers differ between models because each tokenizer has its own voca

Website performance is no longer just a nice-to-have feature—it's a critical factor for user experience, SEO, and business success. Even a one-second delay in page load time can reduce conversions and increase bounce rates. Whether you're building a portfolio, SaaS application, eCommerce platform, or business website, these optimization techniques can make a significant difference. Optimize Images Images are often the largest assets on a webpage. Use modern formats like AVIF or WebP, compress images, and serve responsive image sizes to reduce bandwidth usage. Self-Host Fonts Third-party font requests add latency. Self-hosting fonts, preloading critical font files, and serving only the required character subsets can dramatically improve loading performance. Remove Unused CSS & JavaScript Shipping unnecessary code increases download size and execution time. Tree shaking, code splitting, and removing unused styles help keep your bundle lean. Enable Caching Configure long-term browser caching for static assets and use hashed filenames for cache busting. This allows returning visitors to load your website much faster. Use Lazy Loading Images, videos, and iframes that aren't immediately visible should load only when needed. Native lazy loading is supported by modern browsers and is easy to implement. Optimize Core Web Vitals Google's Core Web Vitals measure how users experience your website. Focus on: Largest Contentful Paint (LCP) Interaction to Next Paint (INP) Cumulative Layout Shift (CLS) Improving these metrics benefits both SEO and user satisfaction. Minify Assets Minify HTML, CSS, and JavaScript files before deployment. Smaller files transfer faster and improve overall performance. Use a CDN Serving assets from edge locations around the world reduces latency and improves loading times for global visitors. Prioritize Accessibility Accessible websites provide a better experience for everyone and often align with SEO best practices. Use semantic HTML, descriptive labe

"Works on my machine" is one of the oldest jokes in software, and it stopped being funny the first time it cost me a weekend. The code was fine. The environment wasn't. A library version on the build box didn't match production, and nobody could see it because "the environment" was a fuzzy, undocumented thing that lived partly in a config management tool, partly in someone's .bashrc , and partly in tribal memory. Containerization is the boring, durable fix for that whole class of problem. Not because containers are magic, but because they force you to turn a fuzzy environment into a single, inspectable, reproducible artifact. That shift — from "a machine we hope is configured right" to "an image we can point at" — is the actual win. Let me walk through what that means operationally, with a minimal example. What containerization actually solves Strip away the tooling and a container image is one thing: your application plus everything it needs to run, packaged together and frozen. The OS libraries, the runtime, the dependencies, your code — all captured at build time into one immutable blob with a content-addressable identity. That has three consequences that matter when you're the one on call: The environment stops being a variable. If it runs from image myapp:1.4.2 in staging, the same image runs in production. You're no longer debugging the difference between two machines. The artifact is immutable. You don't patch a running container in place and hope. You build a new image, tag it, and roll it out. The old one still exists, unchanged, if you need to go back. Rollback becomes trivial. "Roll back" means "run the previous image tag." That's it. No reinstalling packages, no un-applying config drift. After enough years in operations, you learn that most 3 a.m. incidents aren't exotic. They're some version of "this box isn't like the other boxes." Containers don't make you smarter, but they take that entire category off the table. Images vs. containers, briefly These

This is the story of a for-fun project, Luka Fit Index that started with me typing "ai for fun?...

🚀 CNTRL by Omnikon Org Selected for Elite Coders Summer of Code (ECSoC) 2026 Building an AI-first browser for developers—and taking the next step through ECSoC 2026. Open source has always been one of the best ways to learn, collaborate, and build software that makes a difference. Today, I'm excited to share a milestone that means a lot to our team. Our project CNTRL , developed by Omnikon Org , has officially been selected for Elite Coders Summer of Code (ECSoC) 2026 ! 🎉 This selection gives us an incredible opportunity to collaborate with contributors worldwide and continue building a browser that's designed from the ground up for developers. 💡 Why We Started CNTRL Every developer has experienced this workflow: Open documentation Search GitHub Ask an AI assistant Open Stack Overflow Copy code Switch back to the IDE Repeat... The browser has become the center of development, but it still isn't designed for developers. We wanted to change that. Instead of building another browser, we started building one where AI is part of the experience—not another tab. 🌐 What is CNTRL? CNTRL is an AI-powered browser built for developers. Our vision is to create a browser that understands how developers work and helps them stay focused. Some of the ideas we're working toward include: 🤖 AI-assisted coding 📖 Context-aware documentation 💬 Built-in developer assistant ⚡ Faster research workflows 🔌 Extensible architecture 🌍 Community-driven open source The project is still evolving, and ECSoC gives us the perfect platform to accelerate its development. 🏆 Selected for ECSoC 2026 Being selected for Elite Coders Summer of Code 2026 is a huge milestone for our organization. It means we'll have the opportunity to: Collaborate with talented contributors Improve the project's architecture Build exciting new features Learn from the community Grow CNTRL into an even better developer tool We're incredibly thankful to the ECSoC team for believing in our vision. 🌍 About Omnikon Org Omnikon Org is

Introduction Every post in this series has quietly touched a piece of the same problem. Building Agentic Workflows in Java said toolUse.input() is untrusted and must be validated before it reaches your code. Building Reliable LLM Applications in Java said the model will confidently invent facts, so ground it and get typed output instead of parsing prose. Neither post named the thing underneath both statements: anything that crosses from outside your code into the model, or from the model back into your code, is untrusted input — a request body from the network, not a trusted internal value. This post names that boundary directly and gathers the defenses in one place — prompt injection (direct and indirect), input validation, output validation, and PII redaction — with the SAFE pattern shown beside every unsafe one it replaces, since this is the security-forward capstone of the series. The Trust Boundary: Three Kinds of Untrusted Input An LLM application has three places where untrusted text enters: User input — anything a person types, uploads, or submits through an API. Retrieved content — Making RAG Accurate in Java built a pipeline that ranks and returns chunks from a document store; those chunks were written by whoever authored the source document, not by you, and a malicious or compromised document can carry text aimed at the model reading it, not at a human reader. Model output — untrusted the moment it's about to be used rather than displayed : passed to a tool, interpolated into a query, or fed into another LLM call as context. A model that just read attacker-controlled retrieved text can be manipulated into producing attacker-controlled output. The single rule under all three: text is data until your code has explicitly decided it's safe to use for anything more than display. Nothing below is executed against a live API — every snippet is illustrative, and none of it uses a real key or a real record. Direct Prompt Injection: Defending the System Prompt A di

Introduction Every post in this series has quietly touched a piece of the same problem. Building Agentic Workflows in Python said a tool's input is untrusted and must be validated before it reaches your code. Building Reliable LLM Applications in Python said the model will confidently invent facts, so ground it and get typed output instead of parsing prose. Neither post named the thing underneath both statements: anything that crosses from outside your code into the model, or from the model back into your code, is untrusted input — a request body from the network, not a trusted internal value. This post names that boundary directly and gathers the defenses in one place — prompt injection (direct and indirect), input validation, output validation, and PII redaction — with the SAFE pattern shown beside every unsafe one it replaces, since this is the security-forward capstone of the series. The Trust Boundary: Three Kinds of Untrusted Input An LLM application has three places where untrusted text enters: User input — anything a person types, uploads, or submits through an API. Retrieved content — Making RAG Accurate in Python built a pipeline that ranks and returns chunks from a document store; those chunks were written by whoever authored the source document, not by you, and a malicious or compromised document can carry text aimed at the model reading it, not at a human reader. Model output — untrusted the moment it's about to be used rather than displayed : passed to a tool, interpolated into a query, or fed into another LLM call as context. A model that just read attacker-controlled retrieved text can be manipulated into producing attacker-controlled output. The single rule under all three: text is data until your code has explicitly decided it's safe to use for anything more than display. Nothing below is executed against a live API — every snippet is illustrative, and none of it uses a real key or a real record. Direct Prompt Injection: Defending the System Prompt

Introduction We already covered picking the right model tier for the task and caching a large shared prefix in https://pg-blogs.netlify.app/posts/11-building-reliable-llm-apps-in-java/ . Those two lines were the tip of a bigger discipline: LLM cost is not a fixed line item, it's an engineering variable — one you can measure and shrink with the same rigor you'd apply to database query time or container memory. This post goes deeper: how input/output pricing actually works, the exact cache_control shape and how to prove a cache hit rather than assume one, the Batches API for work that isn't latency-sensitive, and model routing — using a cheap model to triage, escalating only the hard cases to a stronger one. The honest framing throughout: measure before you optimize. Every technique here has a cost of its own; applied to the wrong workload, "optimization" makes things slower or more expensive. Token Economics: Why the Prefix Is the Bill Anthropic (like every hosted LLM provider) prices input and output tokens separately, and output is always pricier — the model has to generate output autoregressively, one token informed by all the ones before it, while input can be processed in parallel. Representative pricing from the current model catalog: Model Input Output Claude Opus 4.8 $5.00 / MTok $25.00 / MTok Claude Sonnet 5 $3.00 / MTok $15.00 / MTok Claude Haiku 4.5 $1.00 / MTok $5.00 / MTok Two consequences follow directly: Long system prompts, tool definitions, and RAG context are read on every request , not written once. A 20K-token system prompt sent on every one of 10,000 requests is 200M input tokens — at Opus 4.8 rates, $1,000 before a single output token is generated. The shared prefix , not the user's question, is usually where the money goes. A verbose model wastes money twice — once on the extra output tokens themselves, and again because the next turn's messages history now carries that verbosity forward as input on every subsequent call. Trimming max_tokens an

Introduction https://pg-blogs.netlify.app/posts/10-building-reliable-llm-apps-in-python/ closed with a section on picking the right model per task and caching a shared prefix. That was the entry point into a bigger discipline: LLM spend is an engineering variable, not a fixed bill — one you can measure and reduce with the same rigor you'd apply to query latency or memory footprint. This post goes deeper on four levers: how input/output pricing actually works and why the prefix is usually where the money goes, the exact cache_control shape and how to prove a cache hit instead of assuming one, the Batches API for work that isn't latency-sensitive, and model routing — a cheap model triaging requests and escalating only the hard ones. The throughline is honest: measure before you optimize. Every lever here has its own cost; misapplied, it makes things slower or pricier, not cheaper. Token Economics: Why the Prefix Is the Bill LLM providers price input and output tokens separately, and output always costs more — generation is autoregressive (each token depends on every one before it), while input can be processed in parallel. Representative pricing from the current model catalog: Model Input Output Claude Opus 4.8 $5.00 / MTok $25.00 / MTok Claude Sonnet 5 $3.00 / MTok $15.00 / MTok Claude Haiku 4.5 $1.00 / MTok $5.00 / MTok Two things follow: A long system prompt, tool list, or RAG context is billed as input on every request , not written once. Send a 20K-token system prompt on 10,000 requests and that's 200M input tokens — at Opus 4.8 rates, $1,000 before the model has generated a single output token. The shared prefix , not the user's actual question, is usually the dominant cost. Verbose output costs twice — once directly (more output tokens billed at the higher rate), and again because the next turn's history carries that verbosity forward as input. Asking for concise output and setting a sane max_tokens is a cost control, not just a style choice. This is why the tw

Introduction Building Reliable LLM Applications in Python put it plainly: treat model output as a hypothesis to verify, not a fact to trust. Testing Best Practices in Python put the same discipline in pytest terms: a suite only earns trust by asserting the right things at the right level, unhappy paths included. This post is where those two ideas meet — a pytest assertion either passes or fails against a fixed expected value; an LLM's output is a paragraph of prose that might be right in spirit while differing token-for-token from anything you wrote down in advance. Evaluating it takes a harness, not an assert . That harness has three parts: a golden dataset of representative cases with known-good expected behavior, scoring that turns each case into a pass/fail or a number, and regression testing that runs the harness on every change and fails the build when the score drops. Making RAG Accurate in Python already gave you half of this story — recall@k, precision@k, MRR, nDCG measure whether retrieval found the right chunks. This post measures the other half: whether the generated answer built from those chunks is actually good, which is a genuinely different question a retrieval metric can't answer on its own. Everything below is illustrative, non-executed Python, grounded in the same Anthropic SDK shapes as posts 10/11. The Golden Dataset: Curating Cases, Not Just Inputs A golden dataset is a small, hand-curated set of (input, expected behavior) pairs that represents the ways your application is actually used — not a random sample, and not just the cases that already work. Each case needs enough structure to be scored automatically later: from dataclasses import dataclass , field @dataclass class EvalCase : id : str category : str # "extraction", "qa", "summarization", ... input : str # the prompt/question sent to the system under test expected_exact : str | None = None # non-None only for cases scorable by exact match must_contain : list [ str ] = field ( default_f

Introduction Building Reliable LLM Applications in Java put it plainly: treat model output as a hypothesis to verify, not a fact to trust. Testing Best Practices in Java put the same discipline in JUnit terms: a suite only earns trust by asserting the right things at the right level, unhappy paths included. This post is where those two ideas meet — a JUnit test either passes or fails against a fixed expected value; an LLM's output is a paragraph of prose that might be right in spirit while differing token-for-token from anything you wrote down in advance. Evaluating it takes a harness, not an assertEquals . That harness has three parts: a golden dataset of representative cases with known-good expected behavior, scoring that turns each case into a pass/fail or a number, and regression testing that runs the harness on every change and fails the build when the score drops. Making RAG Accurate in Java already gave you half of this story — recall@k, precision@k, MRR, nDCG measure whether retrieval found the right chunks. This post measures the other half: whether the generated answer built from those chunks is actually good, which is a genuinely different question a retrieval metric can't answer on its own. Everything below is illustrative, non-executed Java, grounded in the same Anthropic Java SDK shapes as posts 10/11. The Golden Dataset: Curating Cases, Not Just Inputs A golden dataset is a small, hand-curated set of (input, expected behavior) pairs that represents the ways your application is actually used — not a random sample, and not just the cases that already work. Each case needs enough structure to be scored automatically later: public record EvalCase ( String id , String category , // "extraction", "qa", "summarization", ... String input , // the prompt/question sent to the system under test String expectedExact , // non-null only for cases scorable by exact/programmatic match List < String > mustContain , // key facts a correct answer must mention (programma

Introduction Every agent needs tools, and every tool needs a way to reach the model. Building Agentic Workflows in Python built that connection by hand — a hand-written JSON schema, a loop that dispatches on block.name . LLM Frameworks vs. the Raw SDK in Python showed LangChain's @tool turning a plain function into that same schema via bind_tools . Both are still bespoke : the tool lives inside one process, wired to one agent, in one language. The Model Context Protocol (MCP) solves a different problem: it standardizes the wire format between an AI application and a tool server, so the server doesn't have to be rewritten per agent, per framework, or per language. This post covers what that buys you, builds a minimal MCP server and a client that consumes it — both on the official Python SDK — and gives an honest answer to when reaching for a protocol is worth it over a direct tool call. The Problem MCP Solves Without a shared protocol, every pairing of agent framework and tool needs its own glue code: a LangChain @tool wrapper, a hand-rolled schema for the raw SDK, a different wrapper again for whatever framework a teammate picks next — an integration per framework, per tool. That's an M×N problem. MCP flattens it to M+N. A server exposes tools, resources, and prompts once, over a standard JSON-RPC protocol. Any host application — Claude Code, Claude Desktop, VS Code, or your own agent — creates an MCP client that speaks that same protocol, regardless of which framework built the host. Write the server once; every MCP-aware host can use it without new integration code. The protocol itself is intentionally boring: JSON-RPC 2.0 messages for lifecycle negotiation, tool discovery, and tool execution. Discovery ( tools/list ) and execution ( tools/call ) are the two calls that matter for this post: // tools/list response (abbreviated) { "jsonrpc" : "2.0" , "id" : 2 , "result" : { "tools" : [ { "name" : "get_account_balance" , "description" : "Look up the balance for an ac

Introduction Every agent needs tools, and every tool needs a way to reach the model. Building Agentic Workflows in Java built that connection by hand — a hand-written Tool schema, a loop that dispatches on toolUse.name() . LLM Frameworks vs. the Raw SDK in Java showed LangChain4j and Spring AI turning an annotated Java method into that same schema via reflection. Both are still bespoke : the tool lives inside one process, wired to one agent, in one language. The Model Context Protocol (MCP) solves a different problem: it standardizes the wire format between an AI application and a tool server, so the server doesn't have to be rewritten per agent, per framework, or per language. This post covers what that buys you, builds a minimal MCP server and a client that consumes it — both on the official Java SDK — and gives an honest answer to when reaching for a protocol is worth it over a direct tool call. The Problem MCP Solves Without a shared protocol, every pairing of agent framework and tool needs its own glue code: a LangChain4j tool wrapper, a Spring AI @Tool method, a hand-rolled schema for the raw SDK — three integrations for one capability, repeated for every tool and every framework you add. That's an M×N integration problem. MCP flattens it to M+N. A server exposes tools, resources, and prompts once, over a standard JSON-RPC protocol. Any host application — Claude Code, Claude Desktop, VS Code, or your own agent — creates an MCP client that speaks that same protocol, regardless of which framework built the host. Write the server once; every MCP-aware host can use it without new integration code. The protocol itself is intentionally boring: JSON-RPC 2.0 messages for lifecycle negotiation, tool discovery, and tool execution. Discovery ( tools/list ) and execution ( tools/call ) are the two calls that matter for this post: // tools/list response (abbreviated) { "jsonrpc" : "2.0" , "id" : 2 , "result" : { "tools" : [ { "name" : "get_account_balance" , "description"