James Ward (Developer Advocate at AWS, Agentic AI Foundation Technical Committee Member) joins Marco to map out the fast-moving landscape of AI agents on the JVM. From MCP and ACP to Spring AI, Embabel, and Ktor — James explains how the JVM ecosystem has not only caught up with Python for building agents, but may have surpassed it. He also introduces SkillsJars (putting agent skills on Maven Central), explains effect-oriented programming and why it supercharges AI coding, and shares how he's been shipping five projects in two months entirely from his phone.

💡 In This Episode

• Why the JVM is no longer second-class for AI agents

• MCP vs ACP vs A2A — when to use which

• Spring AI, Embabel (Rod Johnson), Ktor, LangChain4J compared

• GOAP planning and domain-integrated context engineering

• Agent skills vs MCP servers — and why skills are winning

• SkillsJars: versioned, composable skills on Maven Central

• Testing non-deterministic agents with evals

• Effect-oriented programming and why types matter more than ever

Timestamps: 

(00:00:00) Intro

(00:00:49) Guest intro: James Ward, AWS, and the Agentic AI Foundation

(00:01:37) Are developers now orchestrating AI agents?

(00:02:51) Agent setup, context switching, and review fatigue

(00:05:58) Why typed languages matter more in the AI era

(00:07:14) Scala vs Kotlin vs Java

(00:10:02) What agentic frameworks are and why they matter

(00:14:08) MCP explained

(00:19:42) ACP explained

(00:21:56) How to get started with agent protocols and frameworks

(00:23:43) JVM agent frameworks: Spring AI, Embabel, Koog, and LangChain4j

(00:27:50) AIforJVM.com and building projects with AI

(00:29:55) AI from your phone, dopamine, and productivity

(00:33:14) Testing, evals, orchestration, and reliability in agent systems

(00:41:03) What skills are and where they fit

(00:43:46) SkillsJars and packaging skills for the JVM

(00:49:34) Which AI standards will actually last?

(00:55:35) Effect-oriented programming explained

(01:06:37) Giveaway question

(01:08:24) Rapid-fire round

(01:10:58) Outro

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/ACP0Nx-sW10

Andres Almiray, Java Champion and creator of JReleaser, joins Marco to talk about the realities of open source, release automation, and the evolving Java ecosystem.

They dive into what it really takes to maintain and grow open source projects, why code isn’t the most important part, and how communication, community, and sustainability determine whether projects thrive or die. The conversation also explores release engineering challenges, why most automation setups fail, and how tools like JReleaser simplify software delivery.

💡In this episode:

• Open source realities: burnout, maintenance, and sustainability

• Why code is NOT the most important part of OSS

• JReleaser and release automation in modern workflows

• Common CI/CD and release mistakes developers make

• Maven vs Gradle: trade-offs and real-world experience

• The future of the Java ecosystem and tooling

• AI in open source: PR spam, licensing, and quality concerns

• Advice for newcomers contributing to open source

Timestamps: 

(00:00:00) Intro

(00:00:41) Guest intro + Java journey

(00:01:45) JReleaser: origin, use cases, and adoption

(00:07:49) Software releases and automation best practices

(00:11:31) JReleaser roadmap and release cadence

(00:14:39) Commonhaus, open source sustainability, and succession

(00:20:22) What makes open source projects successful

(00:25:17) Burnout, community management, and prioritization

(00:31:24) Hackergarten and open source collaboration

(00:34:40) Motivation, Java’s evolution, and favorite features

(00:40:44) Maven vs Gradle

(00:44:29) CI/CD, supply chain security, and the future of Java tooling

(00:53:16) AI, licensing, and open source contributions

(01:01:39) Giveaway question

(01:03:25) Rapid-fire round

(01:06:04) Advice for getting started in open source

(01:08:34) Outro

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://www.youtube.com/watch?v=Jts62hWkRO8

Moritz Halbritter, Spring Boot engineer at Broadcom and team lead for start.spring.io (Spring Initializr), joins Marco to talk about the inner workings of the Spring ecosystem and the future of Java performance.

💡 In this episode:

• GraalVM Native Image vs Project Leyden and Java startup performance
• Observability in Spring Boot (logs, metrics, tracing)
• Developer experience improvements: Testcontainers, Docker Compose, SSL hot reload
• AI coding tools, JSpecify nullability, and the future of Spring

⏱️Timecodes:

(00:00) Teaser

(00:53) Meet Moritz Halbretter from the Spring Boot team

(02:23) From school programming to consulting to Spring

(08:14) How Moritz joined the Spring team

(14:20) Spring Native, GraalVM, and Project Leyden

(25:35) Observability, Micrometer, and customer-driven features

(32:08) Developer experience: Docker Compose, Testcontainers, SSL hot reload

(40:35) JSpecify and annotating Spring Boot for nullability

(45:12) start.spring.io and generating Spring projects

(50:45) Using AI in coding, reviews, and open source PRs

(57:30) Where Spring is headed in the next few years

(1:00:08) Favorite languages, Kotlin, and Linux

(1:01:13) Personal projects: solar monitoring, Modbus, and heat pump predictions

(1:08:46) Giveaway and rapid-fire questions

(1:11:32) Outro

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/FUFsul26rgA

Ronald Dehuysser, creator of JobRunr, joins Marco to talk about distributed job scheduling in Java, building a high-throughput background processing framework, and turning an open-source side project into a profitable business.

They dive into what really happens when microservices lack distributed tracing, why dead letter queues can silently lose invoices, how JobRunr scales to thousands of jobs per second, and what it takes to monetize open source in the Java ecosystem. 

💡In this episode:

• Distributed job scheduling and background processing in Java

• JobRunr architecture and high-throughput performance

• Quartz vs modern scheduling approaches

• Retries, exponential backoff, and reliability patterns

• Dead letter queues and observability challenges

• Microservices vs monoliths in enterprise systems

• Monetizing open source and pro licensing models

• Enterprise sales and scaling a developer product

• Burnout, sustainability, and building a team

• AI, LLMs, and the future of junior developers

⏱️Timestamps

(00:00) Teaser(00:48) Who's Ronald Dehuysser and what's JobRunr(01:37) From enterprise dev to freelancing (and switching to .NET)(11:19) Job scheduling pain and birth of JobRunr(16:21) Quitting, COVID, and building the first version(28:48) First customers and monetizing open source(40:13) Big enterprise deal and going full-time(47:16) Burnout, Vipassana, hiring, and building a team(53:20) Sustainability features and the future of JobRunr(56:08) AI, junior developers, the future of coding(01:07:28) Giveaway and Rapid-fire questions

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/9Zgw_0kVFk8

Episode description:Thomas Wuerthinger, Founder and Project Lead of GraalVM and Vice President at Oracle, joins Marco to unpack how modern Java runtimes actually work. They explore why duplicating compilers and garbage collectors across languages is a waste of engineering effort, how GraalVM grew from a research project into production technology, and why Native Image unexpectedly became its most impactful feature.

💡 In this episode:

• GraalVM, Native Image, and Java performance

• JIT vs AOT and predictable runtime behavior

• Polyglot runtimes and shared memory models

• Cloud-native Java, startup time, and memory footprint

• Performance lessons from the One Billion Row Challenge

• Branch misprediction and hardware-level bottlenecks

• Research vs product engineering

• AI in compilers and testing

🕐 Timestamps

(00:00) Intro

(01:09) Meet Thomas Wuerthinger & what GraalVM really is

(03:31) Why duplicating language runtimes is a waste

(06:08) How GraalVM started at Sun Microsystems Labs

(10:26) Writing an interpreter and getting a JIT “for free”

(14:26) Going open source and finding real users

(16:48) Why Native Image took off

(23:02) From research project to real product

(26:56) Native Image, Spring, Quarkus, and Java in the cloud

(35:38) Why JIT performance can be unpredictable

(39:31) When JIT still makes sense

(43:02) Python, JavaScript, and polyglot runtimes

(46:16) Using AI in compilers and testing

(01:04:06) The One Billion Row Challenge

(01:09:50) Branch misprediction and performance surprises

(01:13:26) How to think about performance optimization

(01:25:33) Giveaway question

(01:27:11) Rapid fire and wrap-up

New episodes every other Wednesday. Subscribe for in-depth conversations on software engineering, performance, and developer tools.

🎥 Watch the full episode on YouTube: https://youtu.be/naO1Up63I7Q

Daniel Gräfe, founder of Timing and Cotypist, joins Marco to talk about the realities of indie software development after leaving Big Tech. They discuss turning side projects into a real business, risky rewrites, subscription pricing, and how AI fits into day-to-day coding and support.

Daniel shares lessons from more than a decade of building and maintaining Mac apps, including what worked, what didn’t, and why sustainability matters more than quick wins. A grounded conversation for developers thinking about indie SaaS, side projects, or life beyond a full-time engineering role.

💡 Topics in this episode:

• Leaving Google to go indie

• Side projects vs running a business

• Rewriting a product from scratch

• Subscription pricing decisions

• B2B vs B2C tradeoffs

• Using AI in everyday development

• Long-term maintenance and sustainability

🕑 Timestamps:

(00:00) Intro

(04:08) Timing, CoTypist, and Daniel’s background

(05:30) Early side projects and first Mac apps

(13:16) Getting into Google

(20:02) Promotion, performance, Big Tech reality

(27:33) Quitting Google and going indie

(28:27) Rewriting Timing from scratch

(37:52) Launch day, sales, and burnout

(46:37) Subscriptions and pricing decisions

(54:13) Using AI for coding and support

(01:03:34) Advice for aspiring indie developers

(01:12:10) Agency vs intelligence

(01:13:22) Giveaway + outro

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/t4DkaadHLnI

J. B. Rainsberger, TDD coach and software consultant, joins Marco to challenge common assumptions about software testing. They discuss why JB argues that “integrated tests are a scam,” how testing choices affect design and refactoring, and what teams get wrong when they rely too heavily on end-to-end tests. The conversation draws on decades of real-world experience working with development teams of all sizes.

💡 Topics in this episode:

• What “integrated tests” really are and why they cause problems

• Micro tests, fast feedback, and confidence-driven testing

• When deleting tests is the right decision

• How tests create pressure on system design

• Collaboration and contract tests explained

• Why there is no single testing best practice

• Practical advice for junior and senior developers

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

⏰ Timestamps:

(00:00) Teaser

(01:24) Intro

(02:44) How J. B. got into testing

(06:12) What “integrated tests” actually mean

(10:42) Stop asking “what’s the right test size?”

(13:22) Removing irrelevant details from tests

(15:30) Refactoring, coupling, and deleting tests

(18:08) “There’s no best practice.”

(23:52) Old TDD books and timeless ideas

(26:01) Collaboration and contract tests explained

(33:53) “99.5% of your tests can be micro tests”

(46:21) Do you want to be right or solve the problem?

(01:00:48) Advice for junior developers

(01:19:32) Testing strategy distilled

(01:23:58) Manual QA testing

(01:29:44) Giveaway

(01:33:15) Rapid fire and outro

🎥 Watch the full episode on YouTube: https://youtu.be/j0NjFsb-at8

The first season of The Marco Show has officially come to an end! In this Christmas special episode, we’re announcing all of our giveaway winners and wrapping up the season with a fun Q&A with Marco.

🎁 Giveaway winners:

If you’re one of the winners featured in this episode, please reach out to us at [email protected] and let us know what you’d prefer: a 1-year license to one of our JetBrains products, or a coupon for the JetBrains merch store.

Timestamps:

(00:00) Season Finale Intro & What to Expect

(00:34) LLMs with Jodie Burchell: giveaway winner

(01:08) Software performance with Casey: giveaway winner

(01:39) Hibernate with Gavin King: giveaway winner

(02:15) Flyway with Axel Fontaine: giveaway winner

(02:45) SQL with Torben: giveaway winner

(03:16) Quarkus with Kevin Dubois: giveaway winner

(03:46) Jetty with Ludovic: giveaway winner

(04:18) ByteBuddy with Rafael: giveaway winner

(04:52) How to claim your prize

(05:08) Q&A with Marco

(09:35) Outro and meet the team

Thank you to everyone who watched, listened, and supported the show this season. We truly appreciate it. See you next year!

New episodes every other Wednesday. Subscribe for more developer-focused conversations. 🎥 Watch the full episode on YouTube: https://youtu.be/hZemYWNKa6I

Rafael Winterhalter, creator of ByteBuddy and long-time Java consultant, joins Marco to break down the hidden world of JVM bytecode instrumentation. They discuss how ByteBuddy powers tools like Mockito and Hibernate, why legacy Java still dominates enterprise systems, and what it’s really like maintaining a massive open-source project as a single developer.

💡 Topics in this episode:

• How ByteBuddy works and why frameworks rely on it

• JVM bugs, JIT issues, and surprises from instrumentation

• Supporting legacy Java (5, 6, 8…) in modern environments

• Open-source sustainability and avoiding burnout

• Java modules, unsafe, and evolving with new JDK releases

• Rethinking Java build tools and software engineering complexity

🕑 Timestamps:

(00:00) Intro

(01:10) Rafael’s background & the origin of ByteBuddy

(03:00) What ByteBuddy does and why it matters

(04:32) Replacing CGLIB and early challenges

(07:05) ByteBuddy’s design philosophy

(09:15) Mockito, Hibernate and real-world adoption

(13:14) Open source reality

(15:48) Performance myths and JVM behavior

(18:47) JVM bugs, JIT issues and instrumentation pitfalls

(21:11) Supporting legacy Java in enterprise

(23:56) Testing ByteBuddy across many JDKs

(25:53) Why companies still run Java 5/6

(28:25) Engineering vs economics

(30:39) Modules, unsafe and evolving with Java

(36:12) Maintaining a one-person project

(39:31) Conferences and developer evangelism

(42:02) Consulting vs product engineering

(49:51) Burnout and sustainability

(52:02) Thoughts on AI in software development

(57:13) Rethinking Java build tools

(01:05:07) Build security and dependency risks

(01:10:16) Giveaway

(01:10:48) Rapid fire and outro

New episodes every other Wednesday. Subscribe for more developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/AzfhxgkBL9s

Episode description:

Ludovic Orban (Jetty Maintainer and Performance Engineer at WebTide) joins Marco to break down what “web scale” really means, why most systems don’t need Google-level architecture, and how modern Java servers achieve massive throughput. From Jetty’s design philosophy to multi-core CPU realities, distributed transactions, Terracotta’s wild JVM clustering experiments, and the limits of AI-generated code, this episode dives deep into how high-performance Java is built today.

💡 In this episode:

• What “web scale” actually means for real-world systems

• Jetty vs Tomcat and how Jetty became a fast, modern web server

• CPU evolution, memory bottlenecks, and multi-core performance

• Distributed transactions, XA pitfalls, and compensating patterns

• Terracotta’s clustered JVM: engineering lessons and war stories

• Why developers struggle with performance (and how to learn it)

• AI, coding assistance, and why Ludovic doesn’t trust generated code

⏱️ Timestamps:

(00:00) Teaser

(01:10) Who is Ludovic Orban?

(02:04) What “web scale” really means, and who actually needs it

(03:54) How hardware changes reshaped software performance

(06:28) Cloud, containers, and why performance still matters

(07:28) Jetty vs Tomcat: history, adoption & performance

(10:47) What makes Jetty “fast” in practice

(13:21) How the Jetty team prioritizes performance

(15:10) Recent work: fixing complex HTTP/2 bugs

(16:38) How WebTide supports customers beyond Jetty issues

(17:52) Bitronix: Why Ludovic built his own transaction manager

(20:45) Open-source challenges. The rise and fall of Bitronix

(24:19) Distributed transactions vs compensating transactions

(27:07) Where to learn more: Atomikos and modern approaches

(28:25) Terracotta: clustering JVMs and wild engineering stories

(31:20) What Terracotta taught him about the JVM

(33:48) Real-world Java performance mistakes developers make

(40:22) Why learning performance is so hard

(45:40) Kubernetes, abstraction, and performance visibility

(48:50) Hardware that excites Ludovic: Oxide Computer

(50:42) His take on AI and why he doesn’t trust generated code

(53:30) Lessons from Jetty, Terracotta, Bitronix, and Quartz

(56:10) Rapid-fire questions

(01:01:15) Giveaway and outro

New episodes every other Wednesday. Subscribe for more deep, developer-focused conversations.

🎥 Watch the full episode on YouTube: https://youtu.be/fJvg5zTKHeE