Conference Report #

CodeCrafts 2023 happened in Vienna, again in the Anker Brotfabrik. Like last year, the venue was a great fit: enough space to talk, good stage setup, and a relaxed atmosphere. This time it felt like the conference tried to do everything a bit bigger and a bit better. In many parts it worked, but you could also feel the ambition: more topics, more perspectives, and a stronger “architecture and product” mix.

One clear theme I noticed across multiple talks was DDD-inspired thinking: not only “how do we code this?”, but “what is the domain?”, “where do we put boundaries?”, and “how do teams work with complexity?”. Even when speakers did not explicitly say “DDD”, the mindset was there.

Keynotes and talks that stayed with me #

From Manufacturing Flow to Ecological Effectiveness (Dave Snowden) #

This talk was a strong start because it challenged the way we simplify reality in companies. A line that stayed with me: “You will not see what you do not expect to see.” If your organization only expects simple cause-effect relationships, it will miss signals that the system is changing.

Snowden explained different domains like ordered vs complex vs chaotic thinking (Cynefin). My main takeaway: in complex systems, you cannot fully understand the whole. You work with patterns, you probe, and you learn. He also mentioned conditions for emergence: many elements, rich short-range interactions, and elements not aware of the whole system.

A very practical point was about crises: protect engineers from random interrupts. Create a barrier around the devs and keep the noise away so they can actually fix the problem. Another statement felt painfully real: in over-bureaucratic companies, people learn to defraud the process just to get the job done. Processes should make it easy to do the right thing.

Balancing Coupling in Software Design (Vlad Khononov) #

I had seen a similar version of this talk before, but it still works as a reminder. Coupling is not simply “bad”; the goal is to balance it. Some coupling gives you speed, too much coupling kills change. This topic also connected well to other sessions because boundaries, platforms, and architecture styles all influence coupling.

Platform Engineering: A Bedrock for undistracted Development (Jerzy Kirchner) #

This one was very concrete. He connected platform engineering to cognitive load theory and the idea of reducing extraneous cognitive load. Developers should not spend their brain power on remembering internal infrastructure rules, deployment steps, or “how do I get access to X”.

The Spotify “Golden Path” example was a good framing: not strict governance, not full autonomy chaos, but a paved path that makes the good choice the easy choice. I liked the phrasing that platforms can maximize germane cognitive load: the “useful thinking” for the product and domain, instead of context switching and tool friction.

Technical Implementation of a Design System with Figma (Melanie Freilinger) #

A nice bridge between design and engineering. The most interesting part for me was the practical tooling: using a Figma plugin (Token Transformer) to manage design tokens and transform them for different targets (web, Android, iOS, etc.). The workflow idea of auto-committing to a template repo also sounded like a good way to keep design tokens versioned and close to the code.

How to become an AI toolsmith? (Clemens Helm) #

This talk was less about “AI will replace everything” and more about building useful tools with AI. The takeaway that AI loves examples is simple, but it is still the biggest productivity hack: good inputs, good outputs. I also liked the comparison that AI has similar strengths and weaknesses as humans—so you still need checks, feedback loops, and clear boundaries.

Culture – The Constantly Changing Context (Avraham Poupko) #

A good reminder that culture is not a fixed set of values on a slide. It is context, and context changes. Even with the same people, decisions and behaviors shift when incentives, stress, team size, or company phase changes. For me this connected back to Snowden: you can’t “solve” culture like a technical problem.

How We Process North of 250K Events a Second (Armin Ronacher) #

High-scale talks can sometimes feel far away from day-to-day work, but this one had a lot of real engineering lessons. The core message I took: scale is a system problem. It’s not one magic optimization; it’s architecture, data flow, tooling, and being honest about bottlenecks. Even if you never reach 250k events/sec, the mindset of measuring, simplifying paths, and designing for predictable operations is useful.

Continuous Discovery: Build Products That Customers Want (Matthias Gruber & Thomas Pokorny) #

This talk grounded the conference in product reality: discovery is continuous, not a one-time phase. The message matched the rest of the program well: reduce big-bang planning, learn faster, and keep feedback close to the team.

Meta-modern Software Architecture (Neal Ford) #

Ford’s talk added a broad architecture perspective and vocabulary for what many teams feel right now: we combine “old” and “new” ideas and pick what works. One note I wrote down was “raise stuff out of the domain” (probably about separating concerns and keeping domain logic clean). The bigger point for me was that architecture evolves with constraints, not with fashion.

Organization, vibe, and the afterparty #

The second iteration of CodeCrafts felt more ambitious and more confident. Sometimes that means a bit more noise, but overall the energy was great. The afterparty was easy to enjoy—good for meeting people, continuing discussions, and ending the day without rushing.

Final thoughts #

CodeCrafts 2023 delivered a mix of systems thinking, architecture, platform engineering, product discovery, and practical tooling. What I liked most was that many talks pointed in the same direction: complexity is real, so design teams and systems to learn and adapt, not to pretend everything is predictable.

See you next year in Vienna.