Build in Public: How I Launched a Solo Company

I've always believed one thing: doing is a million times more important than thinking. Lots of people want to start a solo company. They talk about it for a year, bookmark hundreds of articles, buy a few courses — but they never build a single website. My AgentClaw project, from buying the domain to launching with 55 articles, 1992 product data records, and a complete operations system, took just 4 hours total. This article documents the entire process — time spent on each step, tools used, and pitfalls encountered.

Day 1, 11:00 AM — I opened Alibaba Cloud and bought the domain agentclaw.sale. Why .sale? Because it's cheap — 45 yuan per year. .com costs around 70 yuan, but .sale works perfectly fine. Domain registration took about 5 minutes — fill in info, select duration, pay. After payment, I added DNS records on Alibaba Cloud for later configuration. This was the day's only expense — and the entire project's only expense so far.

Why This Topic Matters

11:05 to 11:20 AM — Built the Next.js website. I used a starter template I wrote myself, containing basic page structure, SEO configuration files (next-seo, next-sitemap), Tailwind CSS styling, and article reading logic. Nothing complex — just a basic blog template. Cloned it locally, changed the site name and basic info, initialized a Git repo, and made the first commit. The template is tiny — under 200 lines of core code.

11:20 to 11:25 AM — Deployed to Vercel. Opened Vercel, clicked New Project, selected the GitHub repo, imported, waited for build. Since the template was small, build completed in under 20 seconds. Vercel gave me an agentclaw.vercel.app domain — I tested it immediately: homepage loaded fine, article pages rendered correctly. Then I added a CNAME record in Alibaba Cloud's DNS pointing to Vercel. DNS propagation takes a bit of time, but that didn't stop me from continuing.

11:25 AM — Registered Google AdSense. The process is simple: log in to adsense.google.com with a Google account, fill in your site URL and personal info, submit for review. Google AdSense review typically takes about a week, but registering early means once approved, you just add the ad code. I also registered accounts on Taobao Alliance and JD Alliance — same approach: register first, configure later.

Step 1: Find Your Positioning

11:30 AM — Started batch-generating articles. I already had keyword research and content templates ready, so I jumped straight into AI batch generation. The 55 articles covered various aspects of sports suits: buying guides, brand comparisons, scenario recommendations, fabric education, size selection, washing and care, price analysis, etc. Each article was ~3000 words with data and practical advice. AI generation is fast — about 30 to 40 seconds per article. 55 articles plus formatting took about 2 hours total.

1:30 to 2:00 PM — Imported articles to GitHub repo. Each article is an MDX file in the project's content directory. Each has frontmatter (title, slug, description, tags, etc.) and body Markdown. I wrote a batch processing script to auto-format AI output into standard MDX files. Then git add, git commit, git push — all files pushed at once. Vercel automatically detected the new commit on main branch and triggered build. About 30 seconds later, all 55 articles and 62 pages went live.

2:00 to 3:00 PM — Processed 1992 product data records. I had previously collected sports suit product data from Taobao and JD, stored in CSV files. Now I needed to integrate this data into the site. I designed a simple data page showing price distribution, fabric composition statistics, sales rankings, and other analysis results. I also wrote this data into Feishu's multidimensional tables as baseline operational data. Data import and page development combined took about 1 hour.

Step 2: Build the System

3:00 to 3:30 PM — Built the Feishu operations system. This is the content management system I mentioned. I created a new table in Feishu's multidimensional tables with fields for article ID, title, target keyword, status, publish link, etc. I copied all article metadata from the GitHub article list. I also imported product data into another view of the Feishu table and linked it with articles. This way, when writing new articles, I can directly see existing articles and related product data in the Feishu table.

3:30 to 4:00 PM — Final check. I browsed the entire website thoroughly. Checks included: whether each page loads correctly, whether article links work, whether images load properly, whether SEO configuration is correct (title, description, structured data). I checked several representative pages — no issues found. Then I submitted the sitemap to Google Search Console to ensure Google could quickly discover the new articles.

From 11 AM to 4 PM, minus breaks, actual work time was about 4 hours. In those 4 hours, I completed: domain registration and DNS configuration, Next.js site setup and Vercel deployment, 55 articles generated and published (62 pages total), Google AdSense and CPS affiliate registrations submitted, 1992 product data records processed and displayed, and Feishu operations system setup.

Step 3: Content Output

What would this 4-hour output cost with traditional methods? Let's do the math. 55 articles from a professional writer at 200 yuan each: 11,000 yuan, minimum one week. Website development outsourced: 2,000 to 5,000 yuan, one week. Product data analysis: around 2,000 yuan, several days. Feishu operations system: 1,000 yuan if done professionally. Total in traditional mode: at least 16,000 yuan and two weeks. I did it in 4 hours for 45 yuan.

How did this gap happen? Three reasons. First, AI tools drastically reduced content production cost and time. Second, modern cloud infrastructure (Vercel, GitHub) makes site building and deployment incredibly simple. Third, free SaaS tools like Feishu's multidimensional tables replace systems that previously needed to be built from scratch. Combined, they turned something that required a team and budget into something one person could do in an afternoon.

Of course, Build in Public isn't a one-day thing. Day 1's 4-hour launch was just the start. Operations afterward is the long-term work. Building on the initial 55 articles, I added 20 more in the following two weeks. After a month, total articles reached ~100. Now total articles are 255. Daily operations is just a Feishu table + a git push.

Step 4: Traffic Acquisition

The core philosophy of Build in Public is: share your project and process openly, don't wait until it's perfect. So from Day 1, I've been documenting and sharing every step of AgentClaw. This article itself is part of Build in Public. Three benefits: first, sharing forces you to keep taking action. Second, public data and experiences attract like-minded people. Third, these shared stories become source material for your own site content.

If you're working on a similar project, I recommend trying this "launch on Day 1" approach. Don't wait until your site is perfect. Don't wait until you have 100 articles. Don't chase perfection. Buy a domain, fork a template, write a few articles, and launch. Once you launch, user feedback and data will help you adjust direction. If you stay in "preparation mode" forever, you may never truly start.

Final summary of AgentClaw's Day 1 data: 4 hours total time, 45 yuan total cost (domain), 55 articles / 62 pages produced, 1992 product data records, AdSense and CPS affiliate registrations submitted, Feishu operations system established. The site went live after 4 hours. This is the real speed of a solo company in the AI era. No exaggeration, no fluff — one afternoon, one 45-yuan domain, and you can have your own content site. What's left is just continuous writing, continuous optimization, and continuous sharing.

Long-term Strategy