The $5K Production App
Vibes & Verdicts — Issue #6
THE VIBE
The $5K Production App
People keep asking me how much it cost to build Batesly. They expect a big number. Venture-backed founders talk about burning through hundreds of thousands before launching. Agency quotes for custom legal software start at six figures.
My total spend to get to a production application with encrypted document storage, matter management, trust accounting, billing, role-based access control, audit logging, and a client portal: under $5,000.
Here’s where the money actually went:
AWS infrastructure (~$47/month, ~$280 total so far): RDS PostgreSQL is the biggest line item at around $30/month for the smallest production-grade instance. S3 storage is pennies — I’m storing legal documents, not Netflix videos. App Runner charges for compute time and it’s been negligible. KMS for encryption key management is about a dollar a month.
Replit ($20/month, ~$300 total): My primary development environment. The Pro plan gives me enough compute to run Replit Agent for feature development. This is where the actual building happens.
Replit Overages($800 total): I don’t even know WTF happened, but I got charged $50 extra here and there over the course of several months. I’m too tired to look into it.
Claude API (~$10/month, ~$50 total): Powers the code review workflow (Claude Code). The per-token costs are surprisingly low when you’re using it for focused tasks rather than chatting aimlessly.
Claude Max subscription ($100/month, ~$500 total): For my personal strategy sessions, content creation, and everything that isn’t API-based. Arguably my highest-ROI subscription.
Domain and DNS (~$20/year): batesly.io, Cloudflare for DNS management. Nothing exciting here.
Marketing tools (~$200/month, ~$600 total): Loops for email marketing, Attio for CRM, Apollo for prospecting, Storylane for interactive demos. Most have free tiers that covered me initially.
Everything else (~$350): Lovable for the marketing site, various small subscriptions I’ve tried and either kept or dropped. This bucket includes all the tools I paid for, used once, and canceled.
Grand total: roughly $3,000 depending on how you count the tools I’ve since canceled. Call it $5K to be generous and account for the things I’m forgetting… including the first month of working with Scytale for the SOC 2 Type II compliance.
For context: a single month of iManage for a 25-person firm costs more than my entire build. A one-hour consultation with a Big Four firm’s technology advisory practice costs more than my annual AWS bill.
This is the part where I’m supposed to say “and you can do it too!” And you can — sort of. But here’s the really transparent part: the dollar cost is low because the time cost is high.
I’ve spent hundreds of hours building Batesly. Hours I wasn’t billing clients. Hours on evenings and weekends. Hours learning things the hard way that an experienced engineer would’ve known from the start.
The infrastructure is cheap. The tools are cheap. Your time is not cheap. If you’re a practicing professional building on the side, the real cost is the opportunity cost of what you’re not doing while you’re building. I don’t have a clean answer for that. I just know that for me, the tradeoff has been worth it — and I’ll know for sure one way or the other in the next few months.
What I’d do differently with money: I’d have paid for a UX designer earlier. I spent weeks tweaking interfaces that a professional could’ve gotten right in days. I’d have started SOC 2 compliance prep sooner — it’s not cheap (estimate: $6-10K for the first audit), but enterprise customers ask for it and the timeline is waaay longer than you think. And I’d have set up separate dev/staging/production environments from day one instead of doing everything in production like an unhinged person.
The “you need half a million dollars to build software” narrative was true ten years ago. It’s not true anymore. The barriers to building are lower than they’ve ever been. The barriers to building something good — secure, reliable, well-designed, actually useful — are still real. But this time it’s the knowledge barriers, not financial barriers. And knowledge, unlike venture capital, is available to anyone willing to put in the time.
THE VERDICT
Quick Takes
SOC 2 is a tax on selling to enterprises, not a measure of actual security. Some of the least secure systems I’ve seen have SOC 2 Type II certificates. Some of the most secure are too early-stage to have gone through the audit. The certification tells you a company can follow documented processes consistently. It doesn’t tell you the processes are good. It’s table stakes, not a gold standard.
Every bootstrapped founder should publish their costs. The lack of transparency around what it actually costs to build software benefits exactly one group: the people selling you expensive things. If more founders shared real numbers, more people would realize they could start.
The unsexy advantage of being a lawyer who codes: I read the Terms of Service. When I signed up for Anthropic’s API, I actually read the commercial terms and data processing agreement. I know exactly what happens to data sent through the API (nothing — zero retention, no training). Most founders click “Accept” and hope for the best. Knowing what you’re agreeing to is an underrated superpower.
THE CONFESSIONAL
I Haven’t Made a Dollar Yet
Let’s get the number out there: Batesly’s current revenue is $0 (at the time of writing this… about 5 weeks before it goes live). I hope that by the date you’re reading this, I’ve recouped some money.
I have firms in my pipeline. I’ve done demos. People have said encouraging things. But as of this writing, no one has signed a contract and no money has changed hands.
I’m sharing this because I think the “building in public” space has a honesty problem. People share launch posts and demo screenshots and “10 firms in our pipeline” updates. Nobody shares the part where the pipeline is warm but the bank account is cold.
Here’s what I know: the product is real. The market need is real. The timing feels right. And the next few months will either validate all of that or teach me something expensive. Either way, I’ll tell you about it here.
That’s the deal with this newsletter. The real numbers. The real timeline. Not the curated version. If that sounds useful, stick around. If it sounds terrifying, welcome to my Tuesday nights.
Thanks for reading.
I’m Rachel — a practicing attorney building legal tech through vibe coding. Every week I share what I’m learning, what’s working, and what I probably shouldn’t admit publicly.
Know someone building without a CS degree? Forward this. They’ll thank you… maybe?