> For the complete documentation index, see [llms.txt](https://weekly-essay.mikey.wang/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://weekly-essay.mikey.wang/mikey-wangs-memos.md). # Mikey’s Memos I'm Mikey, a long-term thinker who has grown from software engineer to builder. Here I record my ongoing observations on AI, software engineering, product, business, and personal productivity systems. I care more about the structural changes behind the tools: how technology reshapes organizations, business, and individuals. If you value hands-on experience, independent judgment, and a long-term perspective, you can subscribe via RSS. {% updates format="full" %} {% update date="2026-08-10" %} ## W33 - Five Stages Toward the Future of Work Through the Lens of Agent Teams The maturity of an Agent Team does not depend on the number of agents, but on whether roles, routing, evidence, and permissions form an observable, continuously adjustable work system. Read more: {% endupdate %} {% update date="2026-08-03" %} ## W32 - A Builder's Mac Also Needs a Sweeper A Mac environment needs clear ownership of runtimes and agents, and a Sweeper plus a verifiable loop to continuously detect, converge on, and govern resource drift. Read more: {% endupdate %} {% update date="2026-07-27" %} ## W31 - DeepSeek Through Leaked Materials: Compute, Technology, Financial, and Organizational Constraints A highly likely leaked closed-door DeepSeek discussion exposed constraints in compute, technology, finance, and organization, and also showed how leaks raise the trust cost in fundraising. Read more: {% endupdate %} {% update date="2026-07-24" %} ## W30 - Model × Harness: A Smart Routing Concept The combination of the model and the harness jointly determines the cost and quality of a Coding Agent. The next layer of infrastructure will keep evaluating enterprise private tasks and route requests to the Pareto-optimal configuration. Read more: {% endupdate %} {% update date="2026-07-13" %} ## W29 - The Boundaries of Loop Engineering and RSI Loop Engineering is already a real form of productivity; RSI is still the capability frontier approaching us. The key boundary is whether the loop changes task outcomes, workflows, or the mechanisms that create the next generation of capabilities. Read more: {% endupdate %} {% update date="2026-07-06" %} ## W28 - AI's Repricing of Utility Software AI is lowering the cost of building personalized tools. The moat for utility software will shift from feature coverage to reliably completing full workflows. Read more: {% endupdate %} {% update date="2026-07-01" %} ## W27 - A Builder's Personal Productivity System The key to combining everyday tools into a personal productivity system is not the number of tools, but the closed loop formed between information, tasks, and output. Read more: {% endupdate %} {% update date="2026-05-18" %} ## W21 - The Critical Inflection Point of Software Engineering Software engineering is shifting from "humans write code" to "humans design constraints, and agents complete the implementation," which moves the value of engineering upward. Read more: {% endupdate %} {% update date="2026-05-11" %} ## W20 - Heuristic System and Checkout Strategy Platforms By rebuilding checkout strategy management through a Heuristic System, this piece discusses how complex strategies can gain a unified language, verification, and evolution capability. Read more: {% endupdate %} {% update date="2026-05-04" %} ## W19 - Taste Will Become the New Basic Skill As AI lowers execution costs, the ability to judge what is worth doing and what counts as good will become a scarcer basic skill. Read more: {% endupdate %} {% update date="2026-04-13" %} ## W16 - NoCode, Rapid Iteration, and New Engineering Constraints NoCode and AI accelerate iteration, but they also magnify engineering constraints around architecture, quality, and long-term maintenance. Read more: {% endupdate %} {% update date="2026-03-23" %} ## W13 - Meituan's Q4 Earnings and Why the AI Era Needs a Unified, Efficient Toolchain Starting from Meituan's Q4 earnings and AI engineering practice, this piece discusses why teams need a unified, efficient toolchain. Read more: {% endupdate %} {% update date="2026-03-16" %} ## W12 - D2 Observations: Front-End Vitality in the AI Wave This piece looks at how the role of front-end development is changing in the AI wave, and how front-end capabilities are shifting toward more complete problem solving. Read more: {% endupdate %} {% update date="2026-03-09" %} ## W11 - Twitter's Algorithmic Evolution and Iteration in Source Management Looking at Twitter's algorithmic evolution to discuss source management: the stronger the algorithm, the more important it is to actively filter and verify information sources. Read more: {% endupdate %} {% update date="2026-03-02" %} ## W10 - In the Post-OpenClaw Era: The Coding Singularity Has Arrived, but the New Production Relations Have Not Coding capability has made a leap, but organizational structure, incentives, and production relations have not changed in step. Read more: {% endupdate %} {% update date="2026-03-01" %} ## W09 - A Thinking Framework in Extreme Market Conditions When market sentiment becomes extreme, use company culture, business model, independent thinking, and macro signals to replace noise, so judgment can be reviewed and emotions do not erode the framework. Read more: {% endupdate %} {% update date="2026-01-25" %} ## W04 - What Does AI-Ready Mean? AI-Ready is not about chasing a particular model or tool. It is about building engineering standards that can compound over time and naturally absorb advances in LLMs and infrastructure. Read more: {% endupdate %} {% update date="2026-01-18" %} ## W03 - Lessons from Advanced Troubleshooting For difficult front-end issues, distinguish hidden code defects from uncertain runtime behavior: first use a workaround to contain the risk, then rely on long-term observation to get closer to the truth. Technical explanations are not always complete. Read more: {% endupdate %} {% update date="2026-01-11" %} ## W02 - Rethinking Complex Abstractions AI coding is pushing software engineering from a "scarcity economy" toward an "abundance economy." Complex abstractions may no longer lower costs; the dilemma of Tailwind also shows that the old open-source model of monetizing attention is breaking down. Read more: {% endupdate %} {% endupdates %}