THE SIDE HUSTLE

The Civic Pulse is a portfolio company: a set of nonpartisan media sites that each cover a specific politician — Mamdani, Hochul, AOC — and check what they promised against what they've actually done. Two more products are in development: ReviewPoliticians.com, which extends that coverage to every elected official, and CivicStack, an AI procurement-intelligence tool for sales and GTM teams selling into state and local government.

Landing page:

Five years ago that was a small team and real money — a developer, a designer, an email platform contract, probably an agency somewhere sending invoices. I launched it for under $200, alone, nights and weekends, while working a full-time job. The actual receipt: $20 for the domain, $100 for the first month of the AI subscription that functions as my entire staff, $0 for everything else.

This post is the exact playbook, because "you can just build things now" stays abstract until someone shows you the receipts.

A CONFESSION BEFORE WE DIVE IN

Like most people, I was nervous before I started building with Claude Code. I didn't really know how to connect code to a web server, push to GitHub, open a PR — any of that. After a week of working with it and asking it to explain everything it was doing, all of it became second nature. The learning curve everyone fears is about seven days long.

Because I want as many people as possible to just build, I put together two quick-start guides: one for getting going with Claude Code, and one for connecting Vercel, GitHub, and the rest of the pipeline. Subscribe from this post and both PDFs arrive in your welcome email.

THE PRINCIPLE BEFORE THE STACK

My motto is simple: launch fast, learn fast, pivot fast. Not because speed is impressive, but because your first version is a question, not a statement. You don't actually know what people want from the thing until it's live and strangers are using it. Every week you spend polishing pre-launch is a week of real answers you didn't collect. Ship the ugly version, watch what happens, adjust. The tools below matter only because they collapse the cost of asking.

THE STACK

  • Domain — $20/year. That's what ReviewHochul.com cost me, and it's the only thing here you truly can't get free. Fun fact about naming: when I launched this newsletter, SideDoorNews.com was taken. I bought SideDoorsNews.com and moved on the same day. Perfect name, delayed launch — bad trade.

  • Hosting: Vercel. Free tier. Push code, site's live, done. No servers, no DevOps, no 3 a.m. mysteries.

  • Database: Supabase. Free tier. This is where the promise tracker lives — every promise, source, and status change is a row I can query. If your project has any "track things over time" component, you want a real database, not a spreadsheet you'll outgrow in month two.

  • Newsletter: Beehiiv. Free until you have thousands of subscribers, which is exactly the right pricing model: you pay when it's working.

  • Email sends: Resend. Free tier covers a solo operation's transactional and scheduled sends for a long time.

  • Design: Canva. Free tier for social graphics and templates. Good enough is good enough at this stage.

  • Analytics: PostHog and Microsoft Clarity. Both free at this scale. You cannot learn fast without knowing what readers actually click, scroll past, and abandon — guessing is the expensive part.

  • Payments: Stripe. No monthly fee; they take a cut when money moves. The right cost structure for a thing that might not make money yet.

AI: Claude Max at $100/month, and this is the honest center of the guide. The one real cost, and it's the entire staff. Claude writes and debugs the code, drafts and researches, runs my recurring workflows, and does the work of the analyst, junior dev, and copy editor I don't have — with Cursor in the loop for shipping code. A hundred dollars a month sounds like real money until you price what it replaces: it's the cheapest payroll in the history of media companies. I'm not exaggerating for effect: The Civic Pulse at its current scope does not exist as a solo project without AI doing the leverage work. Learning to direct these tools well is the single highest-return career skill I can point at right now.

TOTAL OUT THE DOOR IN MONTH ONE: ABOUT $120. ONGOING: $100 A MONTH, WHICH IS THE WHOLE PAYROLL. EVERYTHING ELSE RIDES FREE TIERS UNTIL YOU HAVE THE TRACTION THAT MAKES PAYING A GOOD PROBLEM.

WHAT "LAUNCH FAST, LEARN FAST, PIVOT FAST" LOOKS LIKE IN PRACTICE

  1. Launch fast: the first version of everything I've shipped was embarrassing compared to what it became. Nobody remembers, because nobody was watching yet — the early obscurity everyone fears is actually cover.

  2. Learn fast: replies, clicks, and unsubscribes are the truth. Opinions before launch, including your own, are guesses. The tracker, the email cadence, the content mix — all of it has been reshaped by what readers actually did rather than what I predicted.

  3. Pivot fast: when something's not working, the sunk cost is smaller than it feels. The whole point of a sub-$200 launch is that walking away from a bad idea costs you a domain fee, not your savings. Cheap experiments make you braver, and braver compounds.

THE REAL BARRIER

It used to be money and gatekeepers. Now it's just shipping. The stack above is available to anyone with a laptop and $200 — which means the differentiator isn't access anymore; it's whether you actually launch. Most people don't. That's the whole opportunity.

Want the playbook? Subscribe below, follow me on LinkedIn, and your welcome email includes both quick-start guides — Claude Code from zero, and the Vercel + GitHub pipeline. Then reply and tell me: what would you build with $200 and three months of nights and weekends? I answer everything, and I'd genuinely like to know what's stopping you.

— Mike