Selected work
Explore how automation, engineering, education and creative technology become practical work. Start with the area that matters to you.
The bilingual publishing guide and optical network training are available in English. Other project details open in Korean, as marked below.
Discuss a project in English ↗Works
Find the work that connects with yours
Setting a Monthly Spend Cap for AI Automations
A practical guide to capping what an AI automation can spend each month. It separates the limit that only warns from the limit that blocks requests, and shows what the automation should do when a limit is hit.
Before Trusting a Season Score
A DMS.Labs operating note on what a Naver DataLab season score measures and three checks to make before acting on a grade of A: peak timing, the relative baseline, and empty cells when a tool did not run.
One Line for Every Automation Run
A practical guide to writing one line per automation run in a table outside the tool: when it ran, what started it, what it did outside and who checked it, so a wrong result can be traced weeks later.
When the Person Behind the Automation Is Away
A practical guide to writing one card per automation: who owns it, who covers for them, who can stop it and where the access details are kept, so work does not stall when its builder is away.
Did AI Answers Cite Our Site?
A DMS.Labs operating note on sorting questions into three layers and recording Google AI Overviews and ChatGPT citations the same way, separating confirmed cells from unconfirmed ones.
Choosing the First Task to Automate
A practical guide to picking the first task for AI automation. One table compares how often the work repeats, how clear the pass standard is, how long a check takes and what a wrong result would cost.
Choosing Data to Share with AI
A practical guide to selecting inputs for AI exercises, with fictional practice copies, residual identification clues and separate checks for training use and retention.
Acceptance Tests for an AI Skill
Separate revision cases from final checks, define mandatory conditions and check that previously working behavior survives changes.
Why Publication and Records Are Separate
An actual DMS.Labs publication-and-ledger gap: the October 11 repair and inspection code show how to restore the connection without republishing.
When to Restart an AI Automation
A practical guide to checking external results and approval before a retry, with completion states, request identifiers and restart points.
Build a Knowledge Base Your AI Remembers
An AI training guide to building a personal knowledge base with Obsidian and the LLM Wiki method, with a blank starter to download.
Before you ask AI to do the job, write the one-page job sheet
An AI skill-building guide that borrows the agent skill format: a six-part job sheet, tested against a list of real failures.
AI training should start with 'Should I hand this task over?'
A training design guide built on three studies: a task boundary map for deciding what to hand to AI, plus a check two weeks later.
Can Muse and dots replace an existing automation setup?
A documentation-based comparison of 38 active routines: research, publishing, Windows workflows and video production, with migration limits kept explicit.
How to use Jev: setup and costs
A 12-page trial covering structured judgments, Windows setup, budget settings, limitations and evidence-based review.
What to keep in view while Claude Code works
A document-based guide to You should Know: benefits, limits, activation requirements and a bounded trial.
How to hand work to an AI agent
A practical Work guide drawn from OpenAI's dots demo: five habits for delegating to AI and where to draw the line on permissions.
Measuring AI adoption
A practical AI transformation guide to revenue, costs and review time, informed by a KISDI study and its limitations.
Giving AI automation permission in three steps
A practical Work guide to widening an AI automation's permissions in order: read, draft, then act.
Publishing a page in two languages
A practical Work guide to checking Korean and English article bodies, reciprocal language links, photographs and mobile layout on the production domain.
One signal.
Many connections.
Optical Network Training
Hands-on training in fiber-to-the-x (FTTx) network architecture, optical power budgets and fault finding with an optical time-domain reflectometer (OTDR). Written in English for engineers making decisions in the field.
From tasks
to systems.
AI Transformation Design
A practical approach to redesigning work around AI: map the workflow, define human and AI responsibilities, plan handoffs, and measure progress over 90 days.
Workplace Automation Lab
Documented experiments and practical workflows for repetitive work, including email, reports, document preparation, field data and trade documents.
AI Skills & Implementation
Implementation notes on connecting AI to real work, from ComfyUI API integrations and image generation pipelines to tools for organizing knowledge and running AI workflows.
Practical AI Education
Hands-on curricula and examples that help learners use AI in their own work. Programs connect problem definition, workflow design and guided practice, including individual technical consulting.
Dreaming Camera / StoryLens
An AI education project that connects photography, AI, writing and music, helping people explore technology through creative work.