Ben St. John

If an AI prompt could actually build a $1M/month business, the guy selling it for $17 wouldn't be selling it for $17. Cheers 🍻 Anyone thinks otherwise?

17

Why hiring a full dev team is actually the slowest way to build software in 2026 What used to take 5 engineers and a PM, now takes 1 founder and an AI CTO. Here's how: • Architecture decisions: AI compares platforms (e.g. Supabase vs Firebase) and picks what fits your use case. • Coding: Tools like Cursor + Claude Code generate full modules from a comment. Landing pages. Stripe integration. Redis queues. • Debugging: A second agent reviews and fixes the first’s code. Zero hand-holding. • SOPs & Fix Logs: After solving a bug, AI writes your technical documentation - automatically. • Security & Infra: AI flags vulnerabilities, guides Fail2Ban setup, and troubleshoots SSH issues like a sysadmin. No burnout. No salaries. No delays. It’s not theory - I’ve shipped complete SaaS MVPs this way in under 7 days. If you're still thinking in terms of headcount, you're building with blinders on.

5

1 Comment

I built a fully functional saas product on Replit in 24 hours... The exact same product a dev quoted me $12K for.... ChatGPT for instructions and logic, a clean little frontend, plug into Stripe done. I thought I had it mastered. It was working. Well, at least it looked like it was working... The product rendered beautifully. Outputs were solid. Flows were smooth. I was honestly proud of it. Then I had real users test it... Immediately, everything fell apart. Data was getting dropped between tools. Edge cases were breaking flows in ways I didn’t expect. Auth was weird. Error handling didn’t exist. It wasn’t a product. It was a demo that behaved if you treated it gently. That’s when it clicked. These AI no-code tools are incredible for getting from zero to “this could work.” But that last 20% — the part that makes it real, stable, and scalable... If you're not a developer, you just don't know what you don't know. Backend structure, routing, caching, auth, gating, logic loops, data security — all the invisible stuff AI won’t tell you to do unless you ask. The real unlock isn’t “AI instead of devs.” It’s “AI with devs.” Guess what happens when a developer builds with AI... 💥 They know what to ask, what to fix, what to ignore. AI gives them speed. They bring the reliability and the infrastructure... the customization, the performance, and the backend of a real build… but, at 10x the speed and a fraction of the cost. People keep talking about AI taking developers jobs. But what happens when devs start using AI to multiply their output 10×? That’s the part no one’s ready for. That’s not replacement — that’s compounding leverage. I’ve been working with a developer friend of mine out in Pakistan who’s helped me ship product after product — fast, functional, AI built. rinse and repeat. He uses it the way it’s meant to be used: as a multiplier, not a shortcut. If you’re building with AI and living in that all too familiar constant fear the next ONE small change is gonna break everything...again hit me up. happy to make an intro.

25

7 Comments

the zero employee company gets all the headlines. Polsia, $6M ARR, one founder. but the real play is the hybrid. agents handling 80% of functional work, humans making judgment calls on the 20% that compounds. established companies that don't figure this out in the next 18 months are building on sand.

7

2 Comments

What happens when you refuse to pay $3,000 for a JavaScript Gantt chart? You build your own composable UI library. Open source it. And then Vercel acquires it. I just published a conversation with Hayden Bleasel — DX Engineer at Vercel and the creator behind Kibo UI, Nextforge, AI Elements, and Streamdown. Some things that stood out to me: • Every major project he’s built started from personal frustration, not a business plan • He converted 5 separate repos into a Turborepo monorepo in the back of an Uber on the German Autobahn — because he was that frustrated with the deploy cycle • He never charged for his templates while everyone else was selling $149 Next.js boilerplates. His take: give the framework away for free, monetize the hosted convenience layer • Streamdown is now too complex for a human to maintain — it’s fully agent-maintained using Claude Code with skills and unit tests • His mental model for AI + development: it’s what Canva did for design. It doesn’t replace developers, it widens the spectrum of who can build software Whether you’re deep in the Next.js ecosystem or thinking about how open source can shape your career — this one’s worth your time. Video link in the comments 👇 #opensource #nextjs #vercel #webdevelopment #aielements #developerexperience

46

5 Comments

A company finds a WordPress developer. Great portfolio. $35/hr. Done. 6 months later -> a hole in the P&L, a half-done or broken site, and a new search for someone better 🤷‍♂️ I've watched this play out dozens of times over 5 years A company reaches out. Before we even get to discuss what they actually need, the one question comes: "What's your hourly rate?" We answer. Their response? "Too expensive. We have offers for way less" They go with the cheaper option, no offence at all. Problem solved, right? Look, your website is a business asset. Same as any other. It can be a cost center or a revenue driver. The difference isn't the dev hourly rate. It's the approach behind the work. 200 hours at $35/hr is NOT the same as 100 hours at $65/hr. One gives you busywork. The other gives you a business outcome. Before you ask "what's your rate?" ask yourself "what's my ROI?" Because behind every task on your website, there should be a thread that connects to your business goals Share your ROI expectations with the dev team before the work starts The development team you see is just the tip of the iceberg Underneath: page speed -> user experience -> conversion rate -> revenue -> business strategy -> and ultimately... ROI. If your current team can't connect their work to your bottom line, that's not a rate problem. That's a strategy failure. What's the first question you ask when hiring a developer? Rate or result?

35

15 Comments