Back to Events
Meetup #1March 11, 2026Completed

OpenAI Codex Community Meetup - Seoul

How We Actually Build with Codex

Invite-based practitioner meetup · 47 attendees including event staff · Korea Investment Accelerator, Seoul

OpenAI Codex Community Meetup - Seoul was designed for practitioners already applying Codex in real projects. Rather than centering the event on demos, the program focused on concrete workflows, design choices, failure cases, and operating strategies around real use.

Group photo from the OpenAI Codex Community Meetup - Seoul

At a glance

Format

Invite-based

The room was curated from applications rather than filled first-come, first-served.

Attendees

43

The meetup stayed intentionally small to keep discussion density high.

Program

3 talks + OpenAI Q&A

The flow combined the speaker sessions with an OpenAI Q&A and networking joined by KyoungHoon Kim, Jaewon Lee, Wonbae Park, Daeyeol Shim, and Gabriel Chua.

Focus

Workflow / Design / Failure / Ops

The event centered on what it actually takes to work with Codex in practice.

Overview

The meetup was built around a simple premise: there is plenty of AI talk, but far less shared detail on how people are actually building with Codex in production-like settings. The event therefore prioritized operating experience over polished demos.

The program moved from enterprise adoption and practical AX work to harness design for long-running Codex sessions and orchestration extensions through OmX. Each talk approached Codex from a different angle, but all of them stayed grounded in execution, tradeoffs, and real constraints.

The evening closed with an OpenAI talk, Q&A, and networking session joined by KyoungHoon Kim, Jaewon Lee, Wonbae Park, Daeyeol Shim, and Gabriel Chua. The discussion moved naturally from product feedback and workflow pain points to deeper questions about how teams structure Codex inside real engineering systems.

Why It Mattered

Built for active practitioners

The intended audience was not casual curiosity. It was developers already using Codex, teams planning concrete adoption paths, and operators experimenting with long-running sessions and orchestration.

About systems, not just prompts

A defining characteristic of the meetup was its emphasis on harnesses, orchestration layers, and structural design decisions rather than prompt tricks alone.

Operational lessons were front and center

The strongest parts of the discussion came from concrete lessons: what breaks, what scales poorly, what needs additional tooling, and what becomes useful in actual team settings.

Program and Recordings

Below is the core program, along with the recordings that are currently available.

Hyeontae Hwang

AI Adoption with Codex in Enterprise and Public-Sector Contexts

A practical session on bringing Codex into enterprise AX work, including real implementation experience and obstacles.

Watch on YouTube

Goobong Jeong

Effective Harnesses for Long Running Codex

A talk on harness design patterns and lessons learned while making larger Codex tasks complete reliably.

Watch on YouTube

Yechan Heo

OmX: Extending Codex as an Operating Layer

A session on using a tmux-based shell layer to extend Codex with hooks, swarms, and operational workarounds.

Watch on YouTube

Gabriel Chua @OpenAI

Talk & Q&A

An English session joined by KyoungHoon Kim, Jaewon Lee, Wonbae Park, Daeyeol Shim, and Gabriel Chua, focused on direct feedback, usage experience, workflow questions, and product discussion.

Photos

Selected photos from the room, the talks, and the discussion around them.

Group photo after the event
Speaker during a presentation
Speaker explaining a topic during the event
Wide shot of the event room
Attendees focused on the talks
Attendees documenting the event
Audience member asking a question
Panel and Q&A with the OpenAI team
Wide shot of the event room
Wide shot of the event room

Credits

Host
Seowoo Han
Planning
Flatto
Operations
Jiyoung Lee and SpaceY
Photo
Jiyoung Lee
Venue
Korea Investment Accelerator