Every essay, lab report, and book summary you wrote in all 4 years of high school fits into a 200k context window. You could upload your entire education and ask, "When did my writing style actually start to get good?" 📝 #StudentLife #AIHistory
Posts by Markus Eisele
💭 Inspired by @maxandersen.xam.dk I forked my own little 🤖𝗻𝗮𝗻𝗼𝗰𝗼𝗱𝗲 𝗰𝗼𝗱𝗶𝗻𝗴 𝗮𝗴𝗲𝗻𝘁 but using @langchain4j.dev ➕ #gemini
2 approaches:
1⃣ single AI service w/ all tools
2⃣ multi-agent w/ specialized tools
Bonus: Web search & Markdown rendering
glaforge.dev/posts/2026/0...
I wrote this for Java teams using AI coding agents: the real risk is not bad code, it’s too much agency.
Shell access, secrets, MCP tools, autonomous changes. At some point the agent stops being an assistant.
That’s where blast radius matters.
buff.ly/HglVwZq
#Java #AICoding #DevSecOps
REST Data Panache gives you CRUD for free.
Security used to take that away.
Quarkus 3.31 fixes this.
You can now lock down generated endpoints with @PermissionsAllowed directly on the interface. No extra code.
buff.ly/uINRedG
#Java #Quarkus #BackendDev #Security
🎂 Commonhaus Foundation Marks Two Years as Open Source’s ‘Missing Middle’ — and Maintainers Are Taking Notice.
Commonhaus’ 2nd anniversary
www.commonhaus.org/activity/305...
Local models can still fool you.
The Java code looks clean. The actual payload going to the model is often much bigger than you think.
I wrote up how I inspect #Quarkus ↔ Ollama traffic with mitmproxy to see the real JSON, tool overhead, and prompt bloat:
buff.ly/eSqhlqa
#Java #AIEngineering
“The backyard steel of 1958 looked like steel. It was not steel. Today’s backyard AI looks like AI. It is not AI. A TypeScript workflow with hardcoded if-else branches is not an agent. A prompt template behind a REST endpoint is not a model.”
buff.ly/mNgZWFX
Gute Reise!
I don’t think the biggest question with AI coding tools is “which one is best?”
For me it’s: how much control do I keep?
This piece is my current map for working with IDE agents, CLI agents, and app generators without letting autonomy outrun review.
buff.ly/AmBxb2z
#Java #AI #DevTools
Most real-time demos break under real data.
This one doesn’t.
We take raw BTC trades → sliding windows → Bollinger Bands → volatility regimes
All in plain Java with Quarkus + Gatherers4j
buff.ly/biBi3qG
#Java #Quarkus #Streaming
Join me at 11AM ET on April 30 for IBM Dev Day: Bob Edition buff.ly/Bdaof1f
#ibm #ai #developer #free #register
Code is cheap now.
Confidence isn’t.
Agentic SDLC is pushing teams to ship faster than they can understand what they build.
More code ≠ better software.
The real problem? Missing standards for trust.
buff.ly/WrOuAdI
#Java #AI #SoftwareEngineering
I put together a hands-on Quarkus tutorial on full-text vs vector vs hybrid search.
The interesting part is not the three endpoints. It’s seeing where each one fails, and why hybrid is often the safer default.
buff.ly/EwaJ6xy
#Java #Quarkus #AI
Copy-pasting SKILL.md files between projects does not scale.
I wrote a hands-on tutorial on packaging reusable Quarkus / IBM Bob skills with SkillsJars and Maven, so teams can version and share them like normal dependencies.
buff.ly/qKaYqlK
#Java #Quarkus #AI
„If and when it exists, ownership looks like an individual who deeply gives a shit, making a call without waiting for group-consensus. That individual will be right sometimes, and they’ll be wrong other times, and they’ll own it.“
buff.ly/L1Atgrj
#leadership #collaboration
Your Quarkus login flow may be exposing more than you think.
I put together a hands-on tutorial for using PAR with Quarkus OIDC and Keycloak, so OAuth request parameters stay out of the browser URL.
buff.ly/IRIAeNi
#Java #Quarkus #Security
AI can implement API versioning in seconds.
It won’t fix a bad strategy.
Quarkus + IBM Bob + 3 real approaches 👇
buff.ly/GXMG2Ws
#Java #Quarkus #AI #Backend #Dev
“The first 90 percent of the code accounts for the first 90 percent of the development time. The remaining 10 percent of the code accounts for the other 90 percent of the development time.”
AI tools aren’t the problem.
Context is.
Most bad results come from missing repo context + weak prompts.
If you’re a Java dev hitting this wall, this explains why 👇
buff.ly/vhi8z1u
#Java #AI #LLM #DevTools
COBOL on a MacBook → powered by Java.
Using the FFM API, you can connect legacy systems without full rewrites. This is a simple but powerful modernization path.
👉 buff.ly/hymYlLR
#Java #OpenJDK #COBOL #Dev
Apache Polaris is powered by #Quarkus
buff.ly/B2RQ1Wj