BuildInPublic.so

Turn your commits into stories. Build your audience while building your product - no extra work required.

Comment

Inspiration

Here's the thing - everyone's building with AI now. Cursor, Bolt.new, Claude... anyone can ship code in hours. But there's a problem: when everyone can build, how do you stand out?

Brilliant makers ship amazing stuff that nobody ever discovers. They can build 10x faster with AI, but still suck at the one thing that actually matters: building their audience.

We all know we should "build in public" but let's be honest - developers hate marketing, don't know what to share, and when we do post it's usually boring technical updates that get 2 likes.

So I built buildinpublic.so to solve this exact problem.

What it does

buildinpublic.so automatically turns your development work into engaging content that actually builds your audience.

Here's what it does:

🔄 Automatic commit tracking

  • Installs a GitHub Action in your repo that automatically sends commit metadata to our endpoint
  • No setup headaches - the action installs automatically
  • We only see commit messages and metadata, never your actual code

🧠 AI turns commits into content

  • Takes your boring commit messages and turns them into engaging tweets
  • Creates memes that actually relate to your coding experience
  • Analyzes the context to understand what you're actually building

👻 AI Ghostwriter for everything else

  • Write tweets, threads, long-form posts, landing page copy, ads, VSLs
  • Works for both your project and your personal brand
  • Understands your project context and writes accordingly

📖 Brand book creation

  • Scrapes your Twitter profile during onboarding
  • Creates a brand book that captures your writing style and voice
  • Uses this to write content that actually sounds like you

The result: Your commits become content, and you have an AI writer that knows your brand.

How we built it

I started with a solid boilerplate to give the AI good code patterns to work with - basically gave myself a head start. Built the entire backend with Cursor and Sonnet because I wanted to actually learn and understand the core logic. For the frontend and landing page, I used Bolt AI to move fast and iterate on the visual stuff. Honestly, this hybrid approach worked perfectly - I could focus my learning time on the complex backend pieces while still shipping something that looks good. Plus, using different AI tools for different jobs feels like exactly how we should be building in 2025.

Challenges we ran into

I started with vibecoding but now I'm learning how to code. This made everything harder but also forced me to find simple solutions.

Figuring out GitHub Actions - I had no idea how to get commit data at first. Spent way too long learning how GitHub Actions work, but eventually created and open-sourced one that does exactly what we need.

Auto-installing the Action - Users shouldn't have to manually set up GitHub Actions. Figuring out how to automatically install it in their repo was a pain.

Perfecting the prompts - Getting good commit analysis took forever. Most commits are boring ("fix typo", "update readme") so the AI had to be smart about what's worth turning into content.

Onboarding flow - Making the whole setup process not suck while gathering all the context we need (project info, Twitter scraping, etc.) was honestly a nightmare.

Accomplishments that we're proud of

🚀 End-to-end automation - From commit to tweet with zero manual work

🔓 Open-sourced GitHub Action - Other people can use our Action for their own projects

🧠 Smart commit filtering - AI knows which commits are worth turning into content

👻 Dual-mode ghostwriter - Works for both project updates and personal brand content

📖 Automatic brand voice capture - Scrapes your existing tweets to understand how you write

🛠️ Built while learning - Shipped this as a non-technical founder learning to code

What we learned

GitHub Actions are powerful - Once you understand them, they're the perfect way to hook into development workflows.

Prompt engineering is an art - Getting AI to generate good content from commit messages took way more iteration than expected.

Onboarding is everything - If setup is complicated, people bounce. Had to balance gathering context with keeping it simple.

Brand voice matters - Generic AI content is obvious. Scraping existing tweets to understand someone's voice makes a huge difference.

Simple solutions often work better - Instead of complex integrations, sometimes a GitHub Action that just works is the right answer.

What's next for buildinpublic.so

Short term:

  • Better commit analysis to catch more interesting development moments
  • Connect multiple commits into long-form "building in public" updates
  • Project Story feature - Create a beautiful landing page for each project showing the entire journey with commit messages, memes, tweets, and emotional analysis. Basically your project's origin story told through code.

Medium term:

  • Vibemarketing AI Agent - An AI agent for each project that can analyze your codebase, create tailored marketing plans based on your project's USP and target audience, and handle all sorts of marketing tasks
  • Builder Log MCP server - Let developers log every dev session directly from their IDE. This creates a goldmine of context for AI to generate authentic content and build compelling stories

Long term:

  • Turn this into the ultimate software for marketing your vibecoded projects
  • Hyper-personalized for each project and user
  • Make every indie maker's marketing as good as their code

buildinpublic.so makes building your audience as automatic as pushing code. Because when everyone can build with AI, the developers with community are the ones who win.

Built With

Share this project:

Updates