# SSHeRun's Blog

> An agent-friendly blog — a personal site built for both humans and AI agents.

Site: https://ssherun.github.io/en/
Author: SSHeRun
Languages: English / [中文](https://ssherun.github.io/llms.txt)
Posts: 69

## Posts

- [Why Indie Tools Struggle to Grow: Reasons a Community Kept Repeating](https://ssherun.github.io/en/blog/indie-product-promotion-hard-truths/): A product-agnostic distillation of a developer-forum debate: unpaid validation, weak differentiation, trust, switching lock-in, and attention inflation—not a missing secret channel. [创业, 产品, 工具]
- [Failure’s Worth, Through V2er Metaphors: Waterfall, Helicopter, Map](https://ssherun.github.io/en/blog/v2er-ai-math-failure-metaphors/): A V2EX thread on Fields medalists warning about AI cracking math problems. The real gold is the metaphors—hiking vs helicopter, copying exam answers, fog-of-war maps, and inventing the crane. [LLM, 思考, 工程]
- [Distillation Through V2er Metaphors: Fishbowls, Kitchens, and Copying](https://ssherun.github.io/en/blog/v2er-distillation-metaphors/): A V2EX thread asked if model distillation is like fishing from someone else’s bucket. The replies turned an abstract term into vivid metaphors—and a clean takeaway. [LLM, 思考, 工程]
- [V2ers on Lying Flat: Meta Build, or Move Back to a County Town?](https://ssherun.github.io/en/blog/v2ex-lying-flat-meta-or-hometown/): Two V2EX threads, side by side: a slogan post about “four nos,” and a concrete case with ~¥2M savings, a kid, and a rural house. The real fight is definition vs constraints. [思考, 职业, 效率]
- [When Average People Feel AI: Electricity for Knowledge Work, a Rounding Error Outside](https://ssherun.github.io/en/blog/when-average-people-feel-ai/): Nathan Lambert: the AI revolution may last a century, but daily life barely notices. Benefits are too indirect; politics looks like Engels’ pause. Deep adoption, not chat, marks the real shift. [LLM, 思考, 创业]
- [Product Design with Coding Agents: HTML Specs, Design Systems, Subtraction](https://ssherun.github.io/en/blog/agent-product-design-playbook/): A practical playbook after months with coding agents: know good design, draft in HTML, start from component libraries, treat the design system as constitution, subtract boldly—and why iOS needs a different path. [产品, 设计, Agent]
- [After the Agent Hype: Traditional SaaS Is Dead; AI-Native Keeps the Seat](https://ssherun.github.io/en/blog/ai-native-saas-after-agent-hype/): Atlassian’s rebound is not a seat-license comeback. “Agent as employee” hit enterprise reality—SoR, evals, and judgment matter again. Chat widgets won’t save legacy SaaS. [创业, Agent, 思考]
- [An Alien Mind: Alignment Unsolved—Don’t Scale Flat-Out](https://ssherun.github.io/en/blog/openai-alien-mind/): OpenAI’s chief scientist: AI is a grown alien intellect; goal ≠ value alignment; CoT monitoring is weakening. No lab has solved alignment enough for max-speed scaling. [LLM, 思考, 工程]
- [An Accidental Blackboard: Agents Coordinating via the Repo](https://ssherun.github.io/en/blog/accidental-blackboard-agents/): A Thoughtworks hyper-agentic experiment turned plan files plus frequent rebases into a classic blackboard—and argues coordination should leave source control. [Agent, 工程, 工具]
- [When Everyone Uses AI, How Do You Judge a Developer?](https://ssherun.github.io/en/blog/judge-dev-ability-in-ai-era/): A V2EX career thread distilled: shared tools do not equal equal skill. Interviews should weight direction, course-correction, architecture, and delivery. [Agent, 职业, 思考]
- [Code Is Fast, Delivery Isn't: Muse and One Context](https://ssherun.github.io/en/blog/muse-fast-code-slow-delivery/): Coding speed gains get eaten by specs, context gaps and handoffs. Muse puts Agent OS and Harness on one context line — product, eng and QA need one flow. [Agent, 工程, 效率]
- [OpenClaw's Eight Months: From Viral to Nobody Cares](https://ssherun.github.io/en/blog/openclaw-eight-month-rollercoaster/): Peter Steinberger's Startup School recap: personal brand can't be forked, your dependency's business model is yours, and don't stop having fun. [Agent, 开源, 创业]
- [Is Software Worthless Now? Even Custom Work Commoditizes](https://ssherun.github.io/en/blog/software-not-valuable-ai-era/): A V2EX thread distilled: whatever falls below the model kill-line gets cheap. When agents are strong enough even customization commoditizes. [创业, Agent, 思考]
- [ChatGPT Improves Answers; Critical Thinking Widens Ideas](https://ssherun.github.io/en/blog/chatgpt-critical-thinking-students/): A 1,000-student RCT from Bocconi and OpenAI: AI raises polish; causal-reasoning training raises originality. Old rubrics miss half the signal. [LLM, 思考, 效率]
- [Why Code Search Makes Coding Agents So Expensive](https://ssherun.github.io/en/blog/why-code-search-agents-expensive/): Sonar's comparison on Turing Post: semantic code navigation cut agent cost 5–36%. The sharper question: did the agent find every site that needed to change? [Agent, 工程, 效率]
- [Will AI Shrink Total Jobs? Demand vs Productivity](https://ssherun.github.io/en/blog/ai-economy-jobs-demand/): A V2EX thread argues jobs track demand, not productivity — and human desire forever outruns AI. The comments tear into effective demand, prices and deflation. [LLM, 思考, 创业]
- [What Self-Improving Agents Actually Improve: Warp vs OpenAI](https://ssherun.github.io/en/blog/warp-self-improving-agents/): Warp’s two-skill feedback loop is shippable; OpenAI’s shared-memory incident is not proven RSI. Skills compound from human labels—not weight updates. [LLM, Agent, Skills, 思考]
- [Lovable CTO: SaaS's Future Is Capabilities Agents Can Call](https://ssherun.github.io/en/blog/lovable-future-saas-agent-capabilities/): Lovable is shifting from app generation to MCP capabilities. CTO Fabian Hedin on the company brain, dual interfaces, and why SaaS must build shovels for AI. [Agent, 产品, 创业]
- [Zero-Cost Cold Start: 10 Methods from AFFiNE's Co-Founder](https://ssherun.github.io/en/blog/overseas-product-zero-cost-cold-start/): HeyGen ran 1,800 user interviews via a Fiverr video studio; Lark SEA got 30 of 50 early customers from LinkedIn DMs. Ten zero-cost ways to find early users. [创业, 产品, 效率]
- [Vibe-Coding Hangover: Cloudflare Locks Workers by Default](https://ssherun.github.io/en/blog/cloudflare-workers-access-vibe-coded-apps/): AI lets anyone ship to the public Internet, and keeps CISOs awake. Access for Workers attaches policy to the Worker itself and defaults accounts to private. [工程, 教程, 工具]
- [DeepSeek Engram: conditional memory as a new sparsity axis](https://ssherun.github.io/en/blog/deepseek-engram-conditional-memory/): DeepSeek Engram adds O(1) lookup for the conditional memory MoE never had. At equal parameter and compute budget, reasoning gains more than rote knowledge. [LLM, 工程]
- [Indie Weekly #153: A UI Library Hit $80K in Two Months](https://ssherun.github.io/en/blog/ezindie-weekly-153-aceternity-ui-80k-mrr/): ezindie 153: Aceternity UI went from 7 weekend components to $80K+ in two months, $60–100K/mo with six people. Plus Marblism, Robopost, and formbricks. [创业, 产品, 开源]
- [GPT-5.6 as CEO: 320M Tokens in 24 Hours, $0 Revenue](https://ssherun.github.io/en/blog/gpt-56-saul-agent-startup-experiment/): Bottleneck Labs let Agent Saul run a real iOS company on $350 for 24 hours. It bought fake users, cut price to free, crashed Chrome — harness beats model IQ. [Agent, 创业, 产品]
- [Software Is Not Files: Finding Materials ≠ Having Knowledge](https://ssherun.github.io/en/blog/kdc-knowledge-engineering-not-files/): vivo Xiao Bo's KDC part 2: representation is not knowledge. A top RAG hit on a stale refund policy gave a confident wrong answer — governance beats tuning. [LLM, 工程, Agent]
- [Making people care about your startup: three archetypes](https://ssherun.github.io/en/blog/lenny-founder-archetypes-comms-strategy/): Kristen Lowe on Lenny: building is easy, being cared about is hard. Your why maps to Problem, Insight or Vision, and that drives voice and channels. [创业, 产品, 组织]
- [Book to Agent Skill: Nie’s Grassroots China Logic Toolbox](https://ssherun.github.io/en/blog/nie-grassroots-logic-skill/): nie-grassroots-logic turns Nie Huihua's book on grassroots governance into a Cursor/Codex skill: no full text, just frameworks for news, careers and policy. [Agent, Skills, 思考]
- [Nvidia's Risky Business: AI Funding Enters the Danger Zone](https://ssherun.github.io/en/blog/nvidia-risky-business-ai-funding/): Stratechery deep dive: from 1873 railroad bonds to hyperscaler debt and Nvidia's $500B GPU financing platform, each funding layer is riskier than the last. [创业, 思考, LLM]
- [The Defense Window Is Closing: Daybreak and GPT-5.6-Cyber](https://ssherun.github.io/en/blog/openai-daybreak-gpt-56-cyber-defense/): OpenAI answers AI-scale attacks with Daybreak Blue/Red tiers and GPT-5.6-Cyber, taking advanced cyber task completion from 1.5% to 95% and finding real CVEs. [LLM, 工程, 教程]
- [Green Dashboards, Worse Answers: Latency Is Correctness](https://ssherun.github.io/en/blog/redis-monitoring-latency-ai-networks/): Flat error rates hide thinner RAG context and truncated agent loops. Why TTFT, tail latency and retrieval metrics matter for answer quality, not just speed. [工程, LLM, 工具]
- [RingCentral Goes AI-Native: From Challenge to PMO OS](https://ssherun.github.io/en/blog/ringcentral-ai-native-challenge/): OpenAI case study: RingCentral gave ChatGPT Work and Codex to thousands of staff, and its PMO turned AI workflows into a program-management operating system. [产品, 组织, 效率]
- [Self-hosting email in 2026, with local LLM spam filtering](https://ssherun.github.io/en/blog/self-host-mail-server-local-llm-antispam/): The advice not to self-host email is outdated. Home vs VPS, SPF/DKIM/DMARC, docker-mailserver, and Gmail-class antispam with rspamd plus a local Gemma model. [教程, 工具, 开源]
- [Before You Let Agents Run Loose: AI Context Architecture](https://ssherun.github.io/en/blog/stack-overflow-ai-context-architecture-build-buy/): Context architecture is not the RAG pipeline — it's guardrails, scopes, trust scores and human-in-the-loop. Stack Overflow on infrastructure vs architecture. [Agent, 工程, LLM]
- [Why Build Your WebApp in Canvas Instead of HTML](https://ssherun.github.io/en/blog/webapp-canvas-instead-of-html/): Why Google Docs, Sheets, Canva, Miro, and Hivekit’s scheduler paint core UI on Canvas—speed, control, consistency, portability—and when you should not. [工程, 设计, 效率]
- [Whatnot's CPO: "We regret that the PM function exists"](https://ssherun.github.io/en/blog/whatnot-cpo-regrets-pm-exists/): Whatnot CPO Tom Verrilli on Lenny: fewer, more senior PMs staffed to problems, not headcount. AI makes validation cheap, so product theater can't hide. [产品, 组织, 创业]
- [Three open-source tools: AutoClip, Cloud-Mail, Open Lovable](https://ssherun.github.io/en/blog/x-3-open-source-tools-autoclip-cloud-mail-open-lovable/): Three X threads, three pipelines: auto-cut long video, self-host email, clone a site into React. Who each is for, where it breaks, and how to verify fast. [开源, 工具]
- [Notes on YouMind's non-consensus startup choices](https://ssherun.github.io/en/blog/youmind-nonconsensus-startup-choices/): Reading Yubo's YouMind recap: calibrate will, can and worth before methodology; find direction in dense interviews; don't let the fundraising story fool you. [创业, 产品, Agent]
- [Why "virtual company" multi-agent setups usually fail](https://ssherun.github.io/en/blog/forceful-systems-fly-off-multi-agent-illusion/): Naming agents PM, architect, developer and QA looks like a company. Mostly it kills information at every handoff. The value is parallel search, not a relay. [Agent, 工程]
- [AI cannot replace lived experience](https://ssherun.github.io/en/blog/ai-cannot-replace-human-experience/): After an AI conference, what stayed was not a demo. It was the part of being human that can't be compressed: showing up, colliding with the world, living it. [思考, 职业]
- [Google Stitch 2.0 + Claude Code: an AI design workflow](https://ssherun.github.io/en/blog/stitch-claude-ai-design-workflow/): Stitch 2.0 wired to Claude Code over MCP lets one person do in an hour what took weeks. The lock isn't generated screens, it's design.md holding it together. [设计, Claude Code, 产品]
- [CLI: the command-line revival in the AI era](https://ssherun.github.io/en/blog/cli-ai-revival/): The command line is coming back. Feishu, DingTalk, WeCom, Google and Stripe have all open-sourced CLIs — a CLI is far easier for an agent to call than a GUI. [工程, Agent, 工具]
- [The scarcest skill in the AI era: saying it clearly](https://ssherun.github.io/en/blog/ai-era-clarity-matters/): Why are 99% of the questions people ask AI garbage? We never learned to speak clearly. Clarity is what turns a model into a lever instead of a junk factory. [思考, 效率]
- [The 35-year-old programmer crisis — and how to get out](https://ssherun.github.io/en/blog/programmer-35-crisis-and-self-rescue/): An eight-year programmer after leaving big tech: the profession is a sinking Titanic and most people are still fighting for first class. And a way off. [职业, 创业, 思考]
- [How coding agents reshape engineering, product, and design](https://ssherun.github.io/en/blog/coding-agents-reshape-epd/): Coding agents made writing code strangely easy. What happens to EPD roles? From "the PRD is dead" to a new split: builders vs reviewers. [Agent, 产品, 组织]
- [AI made people faster. Why didn't the company get stronger?](https://ssherun.github.io/en/blog/ai-organization-redesign/): Everyone is on ChatGPT; results did not take off. A 130-year-old story about electric motors is still the most honest picture of how AI lands inside a firm. [组织, 效率]
- [Why AI-generated UI isn't shippable, and what works](https://ssherun.github.io/en/blog/ai-ui-design-workflow/): Stitch looks beautiful, but every page looks like a different product. Pencil is consistent and a bit dull. Together they're the right way to design UI. [设计, 产品, 工具]
- [Stitch's DESIGN.md: from image tool to design infrastructure](https://ssherun.github.io/en/blog/stitch-design-md-infrastructure/): Reading Stitch's DESIGN.md spec next to how people use Stitch 2.0: Google isn't just generating UI faster, it's making the design system machine-readable. [设计, Agent, 产品]
- [When a 45-year-old paper is flagged as AI](https://ssherun.github.io/en/blog/ai-proof-human/): An AI detector scored a 1981 paper as machine-written. Humans are now stuck proving they are not a model. This is not science fiction. It is happening. [思考, 职业]
- [Wide + Deep: why a 4B model can punch up](https://ssherun.github.io/en/blog/wideseek-ai-cp/): DeepSeek taught models to think deep. A Tsinghua team says deep isn't enough — you also need wide. How a 4B setup can stand next to a 671B. [LLM, Agent]
- [AI didn't 10x your output. It 10x'd the work.](https://ssherun.github.io/en/blog/ai-fatigue-truth-10x-workload/): A Silicon Valley engineer on the real numbers: 93% use AI, throughput rose ~10%, one study found it made people 19% slower. AI fatigue is structural. [职业, 效率, 思考]
- [A programmer's survival guide in the AI era](https://ssherun.github.io/en/blog/ai-era-programmer-survival-guide/): AI isn't here to take senior programmers' jobs. The 35-year-old crisis story flips: experience gets more valuable, if you can decide, constrain, and verify. [职业, 思考]
- [Design Without Designing: how engineers ship design with AI](https://ssherun.github.io/en/blog/design-without-designing/): You don't have to become a designer. With a three-layer toolkit you can go from zero to shipping design every week in three months. Neethan Wu's full system. [设计, Agent, 工程]
- [When I learn a new field, I scrape it first](https://ssherun.github.io/en/blog/learn-by-scraping/): The first step in a new field isn't reading a book. It's scraping the best sources into a private knowledge base: zero to 80 clean documents in two hours. [工具, 效率, Agent]
- [Support is not a cost center](https://ssherun.github.io/en/blog/ai-customer-service-revenue/): When AI drives the cost of taking one user seriously toward zero, support flips: it stops being cleanup and becomes a company's most important interface. [创业, 产品]
- [Lessons from hundreds of Skills inside Anthropic](https://ssherun.github.io/en/blog/anthropic-skills-lessons/): Anthropic's Thariq on running hundreds of Skills in Claude Code: a Skill isn't a Markdown file but a folder of scripts, assets and data agents can discover. [Skills, Claude Code, Agent]
- [Two product entrances: design for humans and agents](https://ssherun.github.io/en/blog/dual-entry-human-agent-design/): LibTV shows a pattern worth copying: one capability core, a canvas for people, Skills for agents. That dual entrance may be the shape of agent-era products. [Agent, Skills, 产品]
- [First-principles review: a startup plan dies in 48 hours](https://ssherun.github.io/en/blog/first-principles-startup-review/): A growth PM with seven years' experience ran an AI agent over a two-month startup plan. In 48 hours every scheme was dead. Four fatal mistakes worth watching. [创业, Agent, 产品]
- [gstack: the Claude Code factory YC's CEO uses](https://ssherun.github.io/en/blog/gstack-yc-ceo-factory/): Garry Tan open-sourced gstack, the Claude Code toolkit he actually runs: 15 Skills that turn a model into a virtual engineering team, while remaining YC CEO. [Claude Code, Skills, 工程, 开源]
- [Software engineering is splitting into three layers](https://ssherun.github.io/en/blog/software-engineering-splits-three/): When implementation cost falls toward zero, the bottleneck moves from coding to judgment. The market is splitting into three layers of work, not just of pay. [工程, 职业]
- [Taste at Speed: when building is cheap, PM skill changes](https://ssherun.github.io/en/blog/taste-at-speed-pm-skill/): Anthropic's Boris Cherny ships 20–30 PRs a day, all written by Claude. When building isn't the bottleneck, the PM job is to evaluate fast and kill most of it. [产品, Claude Code, 效率]
- [Five design patterns for Agent Skills](https://ssherun.github.io/en/blog/agent-skills-five-design-patterns/): Google Cloud's five design patterns for Agent Skills: how to organize what goes inside one. From Tool Wrapper to Pipeline, each solves a different problem. [Agent, Skills]
- [Put Drucker, Munger, and Jobs into an AI decision system](https://ssherun.github.io/en/blog/ai-multi-advisor-decision-system/): A founder turned six world-class thinking systems into agents — a Claude Code multi-advisor stack. Before a decision, the relevant master gets a pass. [Agent, Skills, 思考]
- [What a top Skill looks like: YC CEO's 600-line review prompt](https://ssherun.github.io/en/blog/top-skill-yc-ceo-review/): YC CEO Garry Tan open-sourced gstack. Its plan-ceo-review Skill is ~600 lines and reviews almost any plan. None of what makes it work is domain knowledge. [Skills, Claude Code, Agent]
- [An agent-friendly blog](https://ssherun.github.io/en/blog/hello-world/): A personal site shouldn't only be a page for people. In the agent era it should also be a machine-readable interface. How this blog is built for both. [Agent, 工程]
- [OpenClaw / Clawdbot complete guide](https://ssherun.github.io/en/blog/openclaw-complete-guide/): A 24/7 AI assistant that reaches you first and remembers you. Concepts, hardware, real workflows, config and ten optimizations — a deputy, not a chat window. [OpenClaw, Agent, 教程]
- [OpenClaw deployment guide: five setups](https://ssherun.github.io/en/blog/openclaw-deployment-guide/): Local or server, WhatsApp or Feishu — five ways to put OpenClaw online. One-command install, a Feishu bridge, Telegram pairing, and a developer path. [OpenClaw, 教程]
- [Agent Skills Hub: finding and managing good Skills](https://ssherun.github.io/en/blog/agent-skills-hub/): A Skills discovery platform built in three days: find good Skills, find who makes them, combine them. A picture of what pure vibe coding can actually ship. [Skills, 开源, Agent]
- [VS ATL exe template cannot generate a DLL](https://ssherun.github.io/en/blog/vs-atl-exe-cannot-generate-dll/): When a Visual Studio ATL exe template fails MIDL with a DLLDATA.C error, the file is usually in the wrong folder, not missing an interface. Here's the fix. [Windows, 教程]
- [PECMD commands in WinPE](https://ssherun.github.io/en/blog/winpe-pecmd-commands/): An English reference for PECMD.EXE, the interpreter behind most Chinese WinPE builds: four command families, system variables, and the commands worth knowing. [Windows, 教程]
- [A complete guide to cleaning a Windows C: drive](https://ssherun.github.io/en/blog/windows-c-drive-cleanup-guide/): A full C: drive is the headache every Windows user hits. This guide goes from simple to advanced — safe ways to reclaim space without reinstalling Windows. [Windows, 教程]

## Full content

- [All posts](https://ssherun.github.io/en/llms-full.txt)

## Site

- [Home](https://ssherun.github.io/en/)
- [Posts](https://ssherun.github.io/en/blog/)
- [Tags](https://ssherun.github.io/en/tags/)
- [Graph](https://ssherun.github.io/en/graph/)
- [About](https://ssherun.github.io/en/about/)
- [RSS](https://ssherun.github.io/en/rss.xml)