Notes Archive
Articles on AI usage, AI agents, knowledge assets, and judgment criteria.
36 articles
An AI note that reframes an insight database as a personal OS that turns judgment criteria, prohibitions, article ideas, and AI instructions into reusable context.
An AI note on treating Codex requests as constraint design, stabilizing implementation quality by giving purpose, scope, prohibitions, and completion criteria.
An AI note on why strength now comes from connecting problem definition, design, AI direction, verification, and operation rather than raw task speed.
An AI note on starting agent adoption from evaluation criteria, context design, permissions, and role boundaries before choosing tools.
An AI note on failure patterns, evaluation, verification, and human-side review design.
A note on designing AI agents around roles, data access, decision authority, and evaluation criteria rather than personality.
A note on turning insight logs into judgment criteria, output rules, and prohibitions that a future personal AI can use.
A note on designing information collection as an automated knowledge pipeline.
Designing Notion as a structured knowledge database that AI can reference and reuse.
Why reusable judgment criteria matter more than simply accumulating information.
How AI shifts web production value toward research, UX design, operations, and outcome design.
Why AI becomes more useful when it reads your own products, customers, failures, constraints, and judgment criteria.
How broad practical experience becomes valuable when coordinating AI, tools, information, and specialists into outcomes.
Why personal data such as judgment criteria, observations, failures, and success patterns should be accumulated before agents become mainstream.
Building reusable AI context through automated information retrieval, structuring, and storage.
Designing AI chatbot experiences around direct answers, value proof, UI limits, and assisted problem solving.
PDP design for AI search, AI shopping, and agent-based discovery, including structured data, reviews, and FAQ design.
Why past logs should be treated as summaries of past judgment criteria, not as a replacement for present judgment.
How company rules, culture, and constraints around AI use can affect experience accumulation and future market value.
Using ChatGPT not as an article generator but as a thinking partner whose conversation logs become article ideas and portfolio material.
How human value concentrates in defining the starting conditions and judging the output as AI accelerates the middle process.
How making, publishing, and accumulating work logs become proof of individual value.
The limits of the idea that adding context always improves accuracy, and why field problems are harder than they look.
How the human role shifts toward judgment, design, and responsibility when AI can create drafts.
The domains where human value remains by staying close to change, context, and specialized practical judgment.
Shifting content operations from making polished pieces to continuously streaming thinking as a byproduct.
How AI reduces tasks rather than work itself, increasing the importance of judgment and process design.
Throwing rough ideas to AI, detecting misalignment, and converging through loops instead of expecting the first answer to be correct.
Why value comes from making excellent open-source software usable rather than building everything from zero.
The difference between compressed AI and structured AI, and why fit with personal thinking style matters.
How work speed changes when people design questions, judge quickly, and turn work into systems.
Turning knowledge into reusable assets through accumulation, structuring, and external publication.
Using AI as a converter that turns thinking notes into publishable content with almost no psychological burden.
How professional knowledge and judgment criteria change when they can be templated and externalized.
Abstracting the decision logic of capable people and moving it into AI as a structure that does not depend on one individual.
The strength of combining multiple skill domains rather than relying on a single replaceable task.