OpenLoop is a Pseudo Lab community project that explores how content storytelling and AI-assisted workflows can bring open-source projects to global audiences.
Throughout the season, we will design, test, and document a repeatable workflow that transforms existing Pseudo Lab project content into English-language content for global discovery — while creating clearer pathways for potential contributors to discover and engage with projects through GitHub.
Project Page: https://pseudo-lab.com/projects/f19e207c-1c3d-426b-b485-d1744655c7a0?tab=home
OpenRel = Open Source Relations
Based in South Korea, Pseudo Lab is home to a growing community of builders, researchers, learners, and open-source contributors exploring AI and emerging technologies.
Across the community, projects continuously generate research, technical insights, experiments, and stories worth sharing. However, bringing these stories to global audiences requires more than translation.
It requires a repeatable process for understanding project context, adapting content for global audiences, maintaining technical accuracy, and connecting content discovery with opportunities to contribute.
OpenLoop explores how this process can be standardized, supported by AI where useful, tested and ultimately reused by other Pseudo Lab projects.
Build a reusable human-AI content workflow that helps Pseudo Lab projects reach global audiences and connect with potential open-source contributors.
Rather than simply creating English content, OpenLoop aims to turn the content-production process itself into a reusable workflow.
Our long-term goal is to enable any Pseudo Lab project to use the OpenLoop workflow when it is ready to reach global audiences — making it easier to publish English content, share project stories, and create new entry points for participation through GitHub.
Build manually. Standardize what works. Automate with purpose.
Throughout the season, we will first establish what good global content looks like, translate those learnings into a standardized process, and then experiment with repository-based AI workflows using Codex or Claude Code.
The goal is not to automate every decision. Instead, we aim to identify repetitive parts of the process that AI can handle effectively while preserving human review for storytelling, technical accuracy, and final publication decisions.
- Review existing Pseudo Lab content and global LinkedIn references
- Explore effective post and visual formats for global audiences
- Establish content criteria and global storytelling approaches
- Develop and compare proposed OpenLoop content formats
- Refine the selected approach into a standardized OpenLoop content format
- Map the end-to-end content process and define common steps, guidelines, inputs, outputs, and QA criteria
- Review example repositories and identify approaches to adopt, adapt, or avoid
- Design the shared OpenLoop repository structure and divide workflow components and responsibilities
- Develop assigned workflow components within the shared OpenLoop repository
- Connect individual components into an end-to-end workflow
- Test the integrated workflow using selected Pseudo Lab project content
- Evaluate where human judgment, templates/rules, AI assistance, or automation are most effective
- Refine the workflow based on test results and remove unnecessary complexity or AI usage
- Document the workflow, guidelines, templates, and supporting materials
- Test whether others can reproduce the workflow using the documentation
- Gather community feedback and apply the refined workflow to real Pseudo Lab projects
By the end of the season, OpenLoop aims to establish a reusable OpenLoop Toolkit consisting of:
- Global Content Guidelines — Principles for adapting Pseudo Lab project stories for global audiences
- Standardized Content Workflow — A repeatable process connecting Blog Translation → LinkedIn Post → LinkedIn Visual
- Human–AI Workflow — A repository-based workflow combining human judgment, templates/rules, AI assistance, and automation where appropriate
- Workflow Documentation — Instructions and templates that allow other Pseudo Lab members to reproduce the process
- QA & Evaluation Framework — Criteria for evaluating accuracy, consistency, efficiency, AI/token usage, and human intervention
- Real Project Outputs — Content produced by applying the OpenLoop workflow to selected Pseudo Lab projects
Ultimately, success means that the OpenLoop workflow can be understood, reproduced, and adapted by Pseudo Lab projects beyond the original team.
Every Monday | 18:50–20:00 KST | ONLINE / OFFLINE
| Week | Stage | Date | Format | Key Activities | Expected Outcome |
|---|---|---|---|---|---|
| W01 | OT | 2026.10.05 | OFFLINE | Project introduction & orientation; introduce the OpenLoop vision, milestones, and expected outcomes | Team alignment |
| W02 | BUILD | 2026.10.12 | ONLINE | Review existing Pseudo Lab assets; research and share LinkedIn content best practices connecting Blog and/or GitHub; identify approaches to global storytelling, content structure, and contribution pathways | Best-practice reference pool |
| W03 | BUILD | 2026.10.19 | ONLINE | Develop a proposed LinkedIn content format; apply each proposed format to the same selected Pseudo Lab project content; share and compare approaches | Individual content prototypes |
| W04 | BREAK | 2026.10.26 | — | Review proposed formats and vote on one approach to develop into the standardized OpenLoop content format | Selected content format |
| W05 | CURATE | 2026.11.02 | OFFLINE | Refine the selected approach into a standardized content format; map the end-to-end process; define common steps, guidelines, inputs, outputs, and QA criteria | Standardized OpenLoop content workflow |
| W06 | CURATE | 2026.11.09 | ONLINE | Review example repositories; identify approaches to adopt, adapt, or avoid; design the shared OpenLoop repository structure; divide workflow components and responsibilities | Repository architecture & responsibility map |
| W07 | DEVELOP | 2026.11.16 | ONLINE | Develop assigned workflow components within the shared repository; document implementation decisions; review components and dependencies with peers | Initial workflow components |
| W08 | DEVELOP | 2026.11.23 | ONLINE | Connect individual components into the first end-to-end OpenLoop workflow; review consistency across inputs, outputs, instructions, terminology, and QA | Integrated OpenLoop workflow |
| W09 | DEVELOP | 2026.11.30 | ONLINE | Test the integrated workflow using selected Pseudo Lab project content; record quality, consistency, processing time, AI/token usage, and human intervention; identify bottlenecks | Workflow test results |
| W10 | SCALE | 2026.12.07 | OFFLINE | Refine the workflow based on test results; identify unnecessary AI usage; finalize the balance between human-led, rule/template-based, AI-assisted, and automated tasks | Refined OpenLoop workflow |
| W11 | SCALE | 2026.12.14 | ONLINE | Document how to use the workflow; complete instructions, guidelines, templates, and supporting materials; define inputs, outputs, human review points, and QA | Workflow documentation |
| W12 | SCALE | 2026.12.21 | ONLINE | Test whether the workflow can be reproduced using the documentation alone; share with Pseudo Lab members and collect feedback | Reproducibility test & community feedback |
| W13 | SCALE | 2026.12.28 | OFFLINE | Apply the refined workflow to selected Pseudo Lab project content; incorporate final feedback; finalize the reusable workflow and documentation; conduct project retrospective | Final OpenLoop Toolkit |
| W14 | BREAK | 2027.01.04 | — | Prepare for final sharing and presentation | Presentation preparation |
| W15 | GRAND GATHERING | 2027.01.09 | OFFLINE | Share the OpenLoop project, workflow, findings, and outcomes with the wider community | Grand Gathering Event |
Build manually. Standardize what works. Automate with purpose.
Blog Translation
↓
LinkedIn Post
↓
LinkedIn Visual
↓
QA & Human Review
↓
Publication
↓
Global Discovery
↓
GitHub Contribution
↻
OpenLoop begins with human-led content experimentation before introducing AI assistance and automation where appropriate.
This allows us to first understand what good content looks like, then determine which parts of the process are best handled through human judgment, templates or rules, AI assistance, or automation.
A core part of OpenLoop is exploring how the standardized content process can be translated into a shared repository-based workflow.
Rather than building separate workflows, participants will work within one shared OpenLoop repository, taking primary responsibility for different workflow components while reviewing and integrating them together.
Before development begins, the team will review example repositories together to understand how instructions, guidelines, inputs, outputs, workflow structures, and QA processes are organized.
For each example, we will identify:
- Adopt — Approaches that can be directly useful to OpenLoop
- Adapt — Useful approaches that need modification for the OpenLoop context
- Avoid — Unnecessary complexity or automation that does not add meaningful value
These findings will inform the shared OpenLoop repository structure and the division of workflow responsibilities.
The integrated workflow will be evaluated across:
| Criterion | What We Evaluate |
|---|---|
| Accuracy | Does the output preserve source facts and technical meaning? |
| Content Quality | Is the content clear and appropriate for global audiences? |
| Consistency | Does the output follow the standardized OpenLoop format? |
| Efficiency | How much time does the workflow require? |
| AI / Token Usage | Is AI usage reasonable relative to the value it provides? |
| Human Intervention | How much manual editing or correction is needed? |
| Reusability | Can the workflow be applied to another project without rebuilding it? |
The results will be used to refine the workflow and determine where human judgment, templates/rules, AI assistance, or automation provide the most effective approach.
OpenLoop encourages all members to contribute across Content Storytelling and Workflow Development.
| Role | Name | Focus |
|---|---|---|
| Builder | @Alice |
Project direction & coordination |
| Runner | @name |
TBD |
| Runner | @name |
TBD |
| Runner | @name |
TBD |
| Runner | @name |
TBD |
| Runner | @name |
TBD |
Workflow responsibilities will be defined collaboratively after the shared content process and repository architecture are established.
Each participant will take primary responsibility for a workflow component while contributing to peer review, testing, and integration across the shared repository.
- Create Together — Share ownership across storytelling, translation, design, workflow development, testing, and documentation.
- Experiment & Learn — Test content formats and workflow approaches, document challenges, and improve each iteration.
- Automate with Purpose — Use AI and automation where they meaningfully improve quality, consistency, efficiency, or scalability.
- Build for Repeatability — Turn successful experiments into simple, documented workflows that can be reused beyond a single season.
OpenLoop welcomes anyone interested in bringing open-source projects to global audiences through content storytelling and AI-assisted workflow development.
We're looking for people with:
- Experience or interest in content curation and storytelling
- Interest in AI-assisted workflows for content and marketing
- Interest in open-source communities and global engagement
- Working proficiency in English
- Experience or interest in visual content design using Canva, Figma, or similar tools
- Familiarity with LinkedIn, Instagram, X, or other social content formats
- Willingness to collaborate, experiment, document learnings, and contribute consistently
No single participant is expected to be an expert across every area. OpenLoop is designed for members to learn by building, testing, and sharing workflows together.
Let's OPEN the LOOP together.
There are many ways to participate in OpenLoop:
- 🧭 Builder — Help shape and coordinate the project
- 🏃 Runner — Build, test, and document content and workflow components
- 👀 Open Participant — Join open sessions, follow experiments, and share feedback
- 💻 Contributor — Improve the OpenLoop workflow or contribute directly to featured Pseudo Lab projects through GitHub
❗️Join the community: Pseudo Lab Discord
❗️Communication channel: Discord #{{channel-name}}
Anyone interested in OpenLoop is welcome to join our open sessions.
You can participate by:
- Joining our regular open sessions through the Pseudo Lab Discord
- Participating during Magical Week
- Meeting the OpenLoop team at Pseudo Lab community events
- Sharing feedback on the OpenLoop workflow and documentation
- Exploring featured projects and contributing directly through GitHub
This section documents the workflows, experiments, outputs, and learnings created throughout OpenLoop.
- 🔄 OpenLoop Workflow:
URL - 📖 Content Guidelines:
URL - 🗂️ Repository Study:
URL - 📋 Templates:
URL - 🧪 Workflow Evaluation:
URL
| Collection | Featured Projects | Platform | Link |
|---|---|---|---|
OpenLoop Collection #1 |
TBD | URL |
|
OpenLoop Collection #2 |
TBD | URL |
| Date | Update | Link |
|---|---|---|
2026.10.26 |
Standardized Content Format Selected | URL |
2026.11.09 |
Repository Architecture Designed | URL |
2026.11.30 |
Integrated OpenLoop Workflow Tested | URL |
2026.12.28 |
OpenLoop Toolkit Finalized | URL |
2027.01.09 |
Grand Gathering | URL |
OpenLoop is developed as part of Pseudo Lab's Open Academy.
This project is made possible by the builders, runners, contributors, and project teams who openly share their work and ideas across the Pseudo Lab community.
Special thanks to everyone helping make Pseudo Lab projects more accessible to the global open-source community. Every contribution opens another path between project discovery and participation — and keeps the loop moving.
Pseudo Lab is a non-profit community focused on advancing machine learning and AI through open collaboration.
Built around the values of Sharing, Motivation, and Collaborative Joy, Pseudo Lab brings together builders, researchers, learners, and contributors to experiment, share knowledge, and create open-source projects together.
This project is licensed under the MIT License.