Episode 27 - AI Multipliers, Event Catalogs, and the New Event Modeling Mascot

Episode Summary

After a one-week hiatus, Martin and Adam return with exciting news: Adam has a new wiener dog puppy named Leela who becomes the official event modeling mascot (wiener dogs are long, just like event modeling diagrams). The conversation covers Martin’s recent in-person workshop in Switzerland, the challenges of organizational adoption, and why AI changes everything about software development - except the need for good specifications.

Main Discussion Points

• The Event Modeling Mascot: Leela the wiener dog is perfect because she’s long like event modeling diagrams and was bred to hunt rabbits (solving rabbit hole problems)
• Organizational Adoption Challenges: Managers approve event modeling until they realize how fundamental the change is, then ask for more proof and case studies
• AI as the Great Equalizer: AI forces everyone to rethink software development, creating a reset moment for adopting better methodologies
• Why Event Catalogs Are Unnecessary: With event modeling, you already have full context - schemas, workflows, and integration points visible at a glance
• The Multiplier Effect: AI multiplies productivity, but only if you change how you specify and implement systems
• Event Modeling is the Fastest Path: Even without AI, using event modeling with code generation beats any manual coding approach
• The Wall in Workshops: Every workshop hits a point where hard questions emerge, and teams want to postpone them to talk about technology instead
• Refactoring is Obsolete: AI can make sweeping changes across files, making traditional refactoring shortcuts less critical
• Neo Vim Dream: The ideal lightweight development environment combining terminal, AI chat, and control-K shortcuts

Key Takeaways

Event modeling serves as your prototype - there’s no need to build code prototypes when you can validate with stakeholders in half an hour. The AI revolution isn’t free; it requires fundamentally changing how you specify systems. Those doing event modeling and event sourcing benefit most from AI because they already have the discipline of clear specifications. The industry is catching up to what event modeling has been saying: specification quality matters more than coding speed. AI doesn’t eliminate the need for understanding - experienced developers who know what they’re solving will always outperform those who just generate code without comprehension.

Memorable Quotes

1. “Event modeling is your prototype. If you have your event model ready, you don’t need to build your prototype. There is absolutely no reason to build a prototype in code.” - Martin
2. “It’s not your business knowledge that goes to production, it’s your developers understanding of it. And there can be quite a huge difference in that.” - Alberto Brandolini quote referenced by Martin
3. “You cannot out code an encoding agent… just focusing on code is a losing game.” - Martin (from recent talk)
4. “Wiener dogs are long like event models are long. So, I think it’s a perfect mascot.” - Adam
5. “I don’t do your job and my job at the same time. I’ll get fired because I’m not programming.” - Adam on why developers couldn’t proofread requirements
6. “If you add new tables if you want your state you need to join them and the more tables you have the more joins you need to make so it doesn’t get simpler.” - Martin
7. “We have people that can code faster than any AI can generate a solution… after so many years we have these templates, we have the standard way of doing a slice.” - Adam

Key Learnings

1. In-person workshops reveal “the wall” - when teams hit hard questions, they want to postpone them and discuss technology instead of continuing to model the business domain
2. Event modeling eliminates the MVP feedback loop because the model itself serves as an accurate prototype that stakeholders can validate immediately
3. The only reason you’d need an event catalog is if you’re not doing event modeling - the model already provides full context with workflows and integration points
4. AI productivity gains aren’t automatic - you need disciplined specification practices like event modeling to truly benefit from the multiplier effect
5. Small, specialized models running locally will become viable for event sourcing work because the problem domain is so formulaic and well-defined
6. TDD and BDD aimed to help developers spot specification problems during implementation, but that’s too late - event modeling catches issues before any code is written
7. The best way to become an expert is immersion - Martin gave three free tech talks in one week to spread event modeling knowledge
8. Traditional refactoring tools from JetBrains are less critical now that AI can make sweeping changes, but the discipline of understanding code structure remains valuable
9. Event sourcing benefits from AI more than traditional development because projections are just another slice - need a search index? Create another projection
10. Cursor’s three AI interaction modes (control-K inline, chat window, terminal assistance) represent the ideal workflow for AI-assisted development