As always, the biggest challenge for a Seoul event was the drive from Suwon to Seoul. It’s like that every single time. But once I got there, I immediately knew it was worth the trip. (Which is probably why the goal should always be ‘go to the gym,’ not ‘work out.’)
I spent the day with four experienced, passionate builders. Getting methods and tips was great, but what I valued more was being able to set ‘this is how they see it’ side by side with ‘this is how I’ve been seeing it.’ You can find tips with a search — you can’t find that.
Three things in particular really landed for me.
Whether it’s a harness or a loop, the concept itself isn’t new.
It’s just something that’s gained popularity and evolved out of necessity. And as foundation models keep getting better, what matters more isn’t clinging to a specific technique but preparing an environment in advance where AI can work well. I felt the same way. (A ‘good environment’ can mean a lot of things, but based on my recent experience, I’d frame it as context — mine or the organization’s — plus a cycle of verifying and rebuilding.)“I’m someone’s agent too. So let’s become the main agent.” I already knew this, but it hits hard every time I hear it!
Recording my own voice and having it organized for me.
Of the three, this is the one I can put to use right away. (This post itself was written that way.)
I’m also leaving the notes I took from the builders’ introductions.
Goobong Jeong — organizer of Ralphthon and leader of Team Attention.
Jeongmin Lee — the opportunity for solo builders, MCP development. It was clear he’s been thinking hard about how a single-person builder can influence an organization (a company). (Checking LinkedIn afterward, I saw a lot of genuinely useful information going up fast.)
Joon Young Park — running a company through Slack, ontology-based system design. I picked up hints on how to add write-back as my own system grows, and how to get my AI to remember and retrieve things well.
Seungjin Kim — building a game world model, betting that world models are a field with real future potential.
What struck me here is that solo builders like these are naturally pushed toward extreme efficiency, and that a coalition of people who share the same interests — like Team Attention — could really help with that. It also occurred to me, somewhat out of nowhere, that becoming an AI-native organization probably requires radical simplification and removing layers.
One more thing I realized while writing this up is that the idea ‘age is just a number’ cuts both ways, not just upward. Being older doesn’t mean you can’t do it, and being younger doesn’t make you any different — especially in the AI era. In a time of such huge change with no fixed answers, the people who use it a lot, think it through, and actually try things are the ones more likely to leverage the moment and achieve something, regardless of age.
Anyway, today’s realization: age really is just a number!
Finally, thank you to Andrew Kim and Sangah Lee at Elev8 for giving me so much insight and experience. I hope you keep creating gatherings like this!
I’ll make the trip up from Suwon again next time!!
#Ralphthon #Ralphthon #Elev8 #TeamAttention #AINative #AIAgent #Contextengineering #SoloBuilder
