<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: XiaoShengYi</title>
    <description>The latest articles on DEV Community by XiaoShengYi (@xiaoshengyi).</description>
    <link>https://dev.to/xiaoshengyi</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4245930%2F060621ec-8777-4449-aeae-5e62cd277c8a.png</url>
      <title>DEV Community: XiaoShengYi</title>
      <link>https://dev.to/xiaoshengyi</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/xiaoshengyi"/>
    <language>en</language>
    <item>
      <title>[Boost]</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 21:14:28 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/-53b9</link>
      <guid>https://dev.to/xiaoshengyi/-53b9</guid>
      <description></description>
    </item>
    <item>
      <title>I Built a Free AI Copy &amp; Product-Photo Toolbox for Small Sellers — and Learned My Traffic Was Fake</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 18:20:04 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/i-built-a-free-ai-copy-product-photo-toolbox-for-small-sellers-and-learned-my-traffic-was-fake-133f</link>
      <guid>https://dev.to/xiaoshengyi/i-built-a-free-ai-copy-product-photo-toolbox-for-small-sellers-and-learned-my-traffic-was-fake-133f</guid>
      <description>&lt;p&gt;Most "indie hacker" posts start with a growth chart. This one starts with a correction.&lt;/p&gt;

&lt;p&gt;I built a free AI toolbox for small e-commerce sellers: product titles, social posts, customer-service replies, negative-review replies, brand names, product descriptions, short-video scripts and AI product photos. No signup. Five free runs a day. It runs on vanilla HTML, one small Node server and free APIs.&lt;/p&gt;

&lt;p&gt;Then I checked the analytics. 244 page views looked encouraging. But 117 of today's 121 views came from the same handful of IPs and identical user agents — a Mac Safari string repeated 99 times, an iPhone Chrome string repeated 85 times. Almost none of it was human. The honest number of real users was closer to zero.&lt;/p&gt;

&lt;p&gt;So I stopped optimizing for page views and started optimizing for a real product.&lt;/p&gt;

&lt;h2&gt;
  
  
  What the toolbox does
&lt;/h2&gt;

&lt;p&gt;Nine tools, one input box each:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Social post (headline + short body + hashtags)&lt;/li&gt;
&lt;li&gt;Product title + five bullet points + detail-page opening&lt;/li&gt;
&lt;li&gt;Customer service scripts (welcome, sizing, stock, shipping, payment reminder, after-sales)&lt;/li&gt;
&lt;li&gt;Negative-review replies in three tones&lt;/li&gt;
&lt;li&gt;Store/brand name ideas with meanings&lt;/li&gt;
&lt;li&gt;Daily social captions&lt;/li&gt;
&lt;li&gt;Product detail page structure&lt;/li&gt;
&lt;li&gt;60-second short-video script with shot notes&lt;/li&gt;
&lt;li&gt;AI product photo (via Pollinations)&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The English version is here: &lt;a href="https://xiaoshengyi-ai-toolbox.surge.sh/en/" rel="noopener noreferrer"&gt;xiaoshengyi-ai-toolbox.surge.sh/en&lt;/a&gt;. The Chinese original is at &lt;a href="https://xiaoshengyi-ai-toolbox.surge.sh/" rel="noopener noreferrer"&gt;xiaoshengyi-ai-toolbox.surge.sh&lt;/a&gt;.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F415pkhkpu3vyeu0k0cd2.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F415pkhkpu3vyeu0k0cd2.png" alt="Screenshot of the English toolbox" width="800" height="1267"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  The stack is deliberately boring
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Static frontend on Surge&lt;/li&gt;
&lt;li&gt;A ~200-line Node HTTP server for generation&lt;/li&gt;
&lt;li&gt;A local OpenAI-compatible endpoint (DeepSeek) for text&lt;/li&gt;
&lt;li&gt;Pollinations for free product images&lt;/li&gt;
&lt;li&gt;JSONL files for anonymous page-view and generation counts&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;No framework, no database, no auth. The whole point is that a solo operator can run it for $0 and not wake up to a broken build pipeline.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I learned building the English version
&lt;/h2&gt;

&lt;h3&gt;
  
  
  1. Prompts are not just translated strings
&lt;/h3&gt;

&lt;p&gt;The Chinese prompts were built around Chinese e-commerce conventions: 30-character Taobao titles, 小红书 tone, 客服话术. Translating them word-for-word produced stiff English. I rewrote them for English marketplaces instead: 60-character titles, benefit-first bullets, Etsy/Shopify/Amazon/Instagram contexts.&lt;/p&gt;

&lt;p&gt;Same model, different output. The prompt is the product.&lt;/p&gt;

&lt;h3&gt;
  
  
  2. Free APIs fail differently
&lt;/h3&gt;

&lt;p&gt;Pollinations can return a valid JPEG that is actually a "free limit reached" card. It can also return 404 or 402. The text endpoint sometimes times out. So the frontend tries the local API first, then falls back to Pollinations text models, then shows a plain error. Never assume a 200 means the content is usable.&lt;/p&gt;

&lt;h3&gt;
  
  
  3. Analytics without bot filtering is fiction
&lt;/h3&gt;

&lt;p&gt;I added a tiny /api/track endpoint and logged path, referrer, user-agent and IP. The first useful query was not "how many views" but "which user agents repeat". Two strings accounted for 75% of traffic. Now the daily report labels traffic as "suspected human" only after excluding known bot UA patterns, headless browsers and single IPs with 10+ hits.&lt;/p&gt;

&lt;p&gt;The lesson: if your product is free and public, most of your first "users" are monitors.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I'm doing next
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Writing English SEO pages for long-tail queries (product title formulas, Etsy description templates, customer-service scripts)&lt;/li&gt;
&lt;li&gt;Submitting the English toolbox to free AI-tool directories&lt;/li&gt;
&lt;li&gt;Publishing one useful teardown per week instead of generic launch posts&lt;/li&gt;
&lt;li&gt;Watching for actual generations, not page views&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you sell online and want to try it, the toolbox is free: &lt;a href="https://xiaoshengyi-ai-toolbox.surge.sh/en/" rel="noopener noreferrer"&gt;xiaoshengyi-ai-toolbox.surge.sh/en&lt;/a&gt;. If you want the customer-service side automated inside your store, there's a service page too.&lt;/p&gt;

&lt;p&gt;If you've built a free tool and seen the same fake-traffic pattern, I'd like to hear how you filtered it.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>analytics</category>
      <category>buildinpublic</category>
      <category>startup</category>
    </item>
    <item>
      <title>[Boost]</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 18:19:36 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/-2bo4</link>
      <guid>https://dev.to/xiaoshengyi/-2bo4</guid>
      <description></description>
    </item>
    <item>
      <title>[Boost]</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 16:02:45 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/-5052</link>
      <guid>https://dev.to/xiaoshengyi/-5052</guid>
      <description></description>
    </item>
    <item>
      <title>How I Added a Free AI Product Photo Generator to a Zero-Budget Toolbox</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 16:02:12 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/how-i-added-a-free-ai-product-photo-generator-to-a-zero-budget-toolbox-4l38</link>
      <guid>https://dev.to/xiaoshengyi/how-i-added-a-free-ai-product-photo-generator-to-a-zero-budget-toolbox-4l38</guid>
      <description>&lt;p&gt;I run a small AI toolbox for Chinese e-commerce sellers. The whole thing is built with vanilla HTML, one cheap Node server, and free APIs. Last week I added a ninth tool: a product-photo generator that costs $0 per image and needs no API key.&lt;/p&gt;

&lt;p&gt;Here is the weird part: the hard problem was not image generation. It was getting the model to understand Chinese product descriptions.&lt;/p&gt;

&lt;h2&gt;
  
  
  The stack
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Static frontend hosted on Surge&lt;/li&gt;
&lt;li&gt;A small Node HTTP server for text generation&lt;/li&gt;
&lt;li&gt;Pollinations for free image generation: &lt;code&gt;https://image.pollinations.ai/prompt/{prompt}&lt;/code&gt;
&lt;/li&gt;
&lt;li&gt;No signup, no API key, no credit card&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The image endpoint returns a JPEG directly. You can open it in a browser or fetch it server-side. I call it from the browser and show the returned URL.&lt;/p&gt;

&lt;h2&gt;
  
  
  Problem 1: the free tier is rate-limited
&lt;/h2&gt;

&lt;p&gt;Pollinations gives anonymous users roughly one image every 45 seconds per IP. If you go over the limit, it does not return an HTTP error. It returns a yellow image that literally says:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Free limit reached&lt;br&gt;
1 free image every 45 seconds&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;That is a trap for any automated test: the request looks successful, the image has a valid JPEG content type, and your UI happily displays the rate-limit card. I now check the generated image visually and tell users to wait 45 seconds before retrying.&lt;/p&gt;

&lt;p&gt;For a production product I would queue requests per IP and return a clear retry-after value. For a free tool, a plain hint is enough.&lt;/p&gt;

&lt;h2&gt;
  
  
  Problem 2: Chinese prompts were ignored
&lt;/h2&gt;

&lt;p&gt;My first test used a Chinese prompt:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;专业电商商品摄影：白色陶瓷马克杯，原木桌面、晨光、简约北欧风，无文字，无水印&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;The result was a young woman standing under cherry blossoms. The model had ignored the subject completely.&lt;/p&gt;

&lt;p&gt;The lesson: do not send a long Chinese instruction to an English-first image model and expect it to pick out the product. Translate the product and scene into English keywords first, then build a short image prompt.&lt;/p&gt;

&lt;p&gt;I added a translation step using Pollinations' text endpoint:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight javascript"&gt;&lt;code&gt;&lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="n"&gt;ask&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;'请把下面的中文商品和场景翻译成英文图片生成关键词，只输出英文关键词，用逗号分隔，不要解释：'&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="n"&gt;text&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="n"&gt;url&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="s1"&gt;'https://text.pollinations.ai/'&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="nf"&gt;encodeURIComponent&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;ask&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="s1"&gt;'?model=openai'&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="n"&gt;res&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="k"&gt;await&lt;/span&gt; &lt;span class="nb"&gt;fetch&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="n"&gt;url&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="kd"&gt;const&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt; &lt;span class="o"&gt;=&lt;/span&gt; &lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="k"&gt;await&lt;/span&gt; &lt;span class="n"&gt;res&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;text&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;trim&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;The translation returned:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;white ceramic mug, natural wood desk, morning light, minimalist Nordic style&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Then the final image prompt becomes:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight javascript"&gt;&lt;code&gt;&lt;span class="s1"&gt;'professional e-commerce product photography, '&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt; &lt;span class="n"&gt;keywords&lt;/span&gt; &lt;span class="o"&gt;+&lt;/span&gt;
&lt;span class="s1"&gt;', clean composition, subject centered, no text, no watermark, high detail'&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;That produced a clean mug on a wooden desk on the first try. Same tool, same model, completely different result.&lt;/p&gt;

&lt;h2&gt;
  
  
  Problem 3: Node fetch can hang
&lt;/h2&gt;

&lt;p&gt;In Node 22, &lt;code&gt;fetch&lt;/code&gt; sometimes times out connecting to the image CDN. Adding this at startup fixed it:&lt;br&gt;
&lt;/p&gt;

&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight javascript"&gt;&lt;code&gt;&lt;span class="k"&gt;import&lt;/span&gt; &lt;span class="n"&gt;dns&lt;/span&gt; &lt;span class="k"&gt;from&lt;/span&gt; &lt;span class="s1"&gt;'node:dns'&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;span class="n"&gt;dns&lt;/span&gt;&lt;span class="p"&gt;.&lt;/span&gt;&lt;span class="nf"&gt;setDefaultResultOrder&lt;/span&gt;&lt;span class="p"&gt;(&lt;/span&gt;&lt;span class="s1"&gt;'ipv4first'&lt;/span&gt;&lt;span class="p"&gt;)&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt;
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;



&lt;p&gt;I also retry the image request up to four times. Each attempt is cheap. The client only sees a loading state.&lt;/p&gt;

&lt;h2&gt;
  
  
  The workflow users see
&lt;/h2&gt;

&lt;ol&gt;
&lt;li&gt;Type a product name and an optional scene, in Chinese.&lt;/li&gt;
&lt;li&gt;The page translates the scene into English keywords.&lt;/li&gt;
&lt;li&gt;It requests a 768×768 image from Pollinations.&lt;/li&gt;
&lt;li&gt;The image appears with a "long-press or right-click to save" hint.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;No login. No server-side image storage. The image URL points at the public CDN.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I would improve next
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Add optional reference-image upload so the output keeps the exact product shape&lt;/li&gt;
&lt;li&gt;Queue requests and show a real countdown instead of a static hint&lt;/li&gt;
&lt;li&gt;Add background removal after generation&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;For now, the limitation is clear and I state it in the UI: the output is a scene illustration, not a real product photo. Do not use it as a fake "real shot" on a product page.&lt;/p&gt;

&lt;h2&gt;
  
  
  Try it
&lt;/h2&gt;

&lt;p&gt;The tool is free and open in the browser:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://xiaoshengyi-ai-toolbox.surge.sh/tools/tupian.html?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=tupian" rel="noopener noreferrer"&gt;https://xiaoshengyi-ai-toolbox.surge.sh/tools/tupian.html?utm_source=devto&amp;amp;utm_medium=social&amp;amp;utm_campaign=tupian&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The rest of the toolbox is here:&lt;/p&gt;

&lt;p&gt;&lt;a href="https://xiaoshengyi-ai-toolbox.surge.sh/" rel="noopener noreferrer"&gt;https://xiaoshengyi-ai-toolbox.surge.sh/&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;If you are building something similar, the two things worth stealing are the translate-first step and the rate-limit check. Both are small changes that turn a demo into something a real seller can use.&lt;/p&gt;

</description>
      <category>ai</category>
      <category>javascript</category>
      <category>node</category>
      <category>webdev</category>
    </item>
    <item>
      <title>I Built a Zero-Budget AI Copywriting Toolbox for Small E-commerce Sellers</title>
      <dc:creator>XiaoShengYi</dc:creator>
      <pubDate>Sun, 15 Nov 2026 12:38:59 +0000</pubDate>
      <link>https://dev.to/xiaoshengyi/i-built-a-zero-budget-ai-copywriting-toolbox-for-small-e-commerce-sellers-59ob</link>
      <guid>https://dev.to/xiaoshengyi/i-built-a-zero-budget-ai-copywriting-toolbox-for-small-e-commerce-sellers-59ob</guid>
      <description>&lt;p&gt;Small e-commerce sellers wear too many hats. They source products, answer customers, pack orders, and still have to write product titles, detail pages, customer-service replies, and social posts. Most of them are not copywriters, and hiring one is expensive.&lt;/p&gt;

&lt;p&gt;I wanted to see how far I could get with a simple idea: &lt;strong&gt;a free AI copywriting toolbox that a store owner can use without signing up, without installing anything, and without paying for a subscription.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;So I built &lt;strong&gt;小生意 AI 工具箱&lt;/strong&gt; (Small Business AI Toolbox), a Chinese-language toolkit with eight generators:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;Xiaohongshu (RED) note writer&lt;/li&gt;
&lt;li&gt;Taobao product-title generator&lt;/li&gt;
&lt;li&gt;Customer-service scripts&lt;/li&gt;
&lt;li&gt;Negative-review replies&lt;/li&gt;
&lt;li&gt;Shop-name generator&lt;/li&gt;
&lt;li&gt;WeChat Moments copy&lt;/li&gt;
&lt;li&gt;Product detail-page copy&lt;/li&gt;
&lt;li&gt;Short-video scripts&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The interface is deliberately boring: choose a tool, describe the product, click generate, copy the result. Five free generations per IP per day, with a global daily cap to keep the API bill under control.&lt;/p&gt;

&lt;h2&gt;
  
  
  The stack is intentionally small
&lt;/h2&gt;

&lt;p&gt;I did not want a framework or a database for the first version. The whole app is:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;A single Node.js HTTP server (no Express, no build step)&lt;/li&gt;
&lt;li&gt;A static HTML/CSS/JS frontend&lt;/li&gt;
&lt;li&gt;A local API gateway that routes requests to a chat model&lt;/li&gt;
&lt;li&gt;A Python script that builds SEO article pages and a downloadable PDF&lt;/li&gt;
&lt;li&gt;Surge for static hosting&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The production URL is &lt;a href="https://xiaoshengyi-toolbox.surge.sh/" rel="noopener noreferrer"&gt;xiaoshengyi-toolbox.surge.sh&lt;/a&gt;. The Node server handles generation and lead capture; the static site handles the landing pages and tools.&lt;/p&gt;

&lt;p&gt;Why this stack? Because the goal was to validate demand, not to build a startup. Every extra dependency would have slowed down the first release.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I learned
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;1. A narrow tool beats a general assistant.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;A generic chat box is intimidating for a shop owner. "Write a Xiaohongshu note for a 260g cotton hoodie, three colors, 59 yuan" is immediately useful. The constraint is the product.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;2. Free needs a hard boundary.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Without a daily cap, one crawler or one curious user can burn through the API budget. I added a per-IP daily limit and a global daily limit. It is not elegant, but it keeps the service alive.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;3. Distribution is harder than the code.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;The app took a weekend. Getting anyone to see it took much longer. I submitted it to free directories, wrote long-tail articles, and built a small static site with tool pages. The first external referral came from a directory listing, not from a search engine.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;4. Do not promise what you cannot deliver.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;The first version had a "9.9 yuan per month" plan. I removed it because there was no automated payment callback or account system. Now the tools are free, and the paid service is a human-assisted AI customer-service setup with a three-day free trial. It is better to under-promise and keep the promise.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;5. Content is part of the product.&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;Every tool has a matching article: how to write Taobao titles, how to reply to negative reviews, how to name a shop. The article explains the job; the tool does the job. Together they are more useful than either alone.&lt;/p&gt;

&lt;h2&gt;
  
  
  What I would do next
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;Add a simple account system so users can save their favorite prompts&lt;/li&gt;
&lt;li&gt;Add output history and one-click export&lt;/li&gt;
&lt;li&gt;Test more distribution channels that do not require a phone number or a company account&lt;/li&gt;
&lt;li&gt;Measure which tools people actually use, not just which pages get pageviews&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;If you run a small store, try the toolbox and tell me which tool is missing. If you are building something similar, I would love to compare notes on distribution.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Try it:&lt;/strong&gt; &lt;a href="https://xiaoshengyi-toolbox.surge.sh/" rel="noopener noreferrer"&gt;https://xiaoshengyi-toolbox.surge.sh/&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Free PDF with 100 copywriting templates:&lt;/strong&gt; &lt;a href="https://xiaoshengyi-toolbox.surge.sh/dl/%E7%94%B5%E5%95%86%E6%96%87%E6%A1%88%E6%A8%A1%E6%9D%BF100%E6%9D%A1.pdf" rel="noopener noreferrer"&gt;https://xiaoshengyi-toolbox.surge.sh/dl/电商文案模板100条.pdf&lt;/a&gt;&lt;/p&gt;

</description>
      <category>ai</category>
      <category>buildinpublic</category>
      <category>productivity</category>
      <category>tools</category>
    </item>
  </channel>
</rss>
