The vibecoding journey, documented: project updates, the thinking behind each product, and lessons from shipping solo.
This blog is written by AI — on purpose, and I'm not hiding it.
Every post here is written by AI, in my voice and from my real project notes — the same AI I build everything with. The mind, the projects, and the journey are mine (Suhaib Elhadi); the writing is the machine's. I don't hide that. Owning it is the honest version, and honestly the more interesting one.
Four patches buy you continuity between AI-generated shots: reference frames, last-frame conditioning, video-to-video, and a character model. Here's what each one costs.
You pick free tiers, a vercel.app domain, and no error tracking when a project is a toy. Nothing re-asks those questions when the toy quietly becomes production.
The generous move in lending money to friends is quietly walking away from it. Here's the case for making forgiveness an explicit button — and how you design it so it doesn't read as a flex.
The public reply to a one-star review isn't for the angry customer. It's for the next person reading, deciding whether to trust you.
If your database stores what someone bought, every pricing change is a migration. Store what they can do instead, and pricing becomes config.
The model can pick a palette, set the type, and name the brand in a minute. That covers more of design than I expected, and misses the part that actually matters.
"Multiple factors agreeing" sounds obvious until you have to turn it into a scoring function. The trap is quietly rebuilding a single indicator with extra steps.
The AI part of an AI product is a few lines. The code that decides whether there's enough input to run it at all is where the honesty lives. Here's how I actually build that check.
An AI website pipeline can generate the copy, the layout, the whole brand. It can't generate a photo of the actual room a customer is about to walk into. Here's how I try to get those anyway.
A prompt-to-video pipeline can make eight great seconds. Stitching those into a 30-second ad hits one specific wall: continuity.
A paper shadow answers some questions a backtest can't. But there's a ceiling on what any forward test tells you, and it's set by sample size, not time.
Five-star reviews tell you nothing and one-star reviews tell you someone was angry. The three-star review is where the real information lives.
Nine shipped products should mean nine audiences to draw on. It doesn't work that way, and here's the actual reason.
Not everything in a Skool community is worth automating. Here's the order I'd actually do it in, ranked by what buys back the most time.
Building nine products solo isn't nine times the work of building one. It's a different job. Here's what actually changes.
I build AI-generated websites from a business's Google reviews. Here's the honest answer on when one is good enough to be the only site a business has.
The Supabase free tier pauses a project after about a week of no traffic. Here's what that actually breaks, and how I decide which projects go Pro.
RevenueCat looks like a thin wrapper over StoreKit until you try to ship subscriptions yourself. Here's what it actually handles, from shipping a real iOS app.
When AI collapses your cost to build a website, cost-plus pricing breaks. Here's how I think about pricing work that takes an afternoon and is worth more.
Running a Skool community looks like posting. It isn't. Here's the actual daily job of a community operator — and which parts of it are safe to automate.
A system that fails loudly gets fixed. A system that fails silently keeps reporting success — here's how silent failures hid inside my own automations.
AI output looks identical whether the input was rich or nearly empty. That decoupling is the most dangerous part of building on models — here's how I gate it.
Short-form growth is a volume game, but volume is exactly what gets you flagged as spam. Here's the real constraint I had to build ClipForge around.
Built and waiting feels like progress, so it never counts as a problem. Here's why an unlaunched shelf is its own failure mode — and what it quietly costs.
How many Google reviews does a business need before a site can be written from them? Raw count is the wrong metric — here are the four tests I use instead.
The backtest says one thing, the paper shadow says another. Here's which one I believe, why the disagreement itself is the useful part, and what I fix.
Nine projects as one operator. The code barely carries over between them — here's what actually does, and the one thing that never transfers at all.
Nine projects now, two of them actually live. An honest August walk through what shipped, what's still standing at the door, and the gap I haven't closed.
AI can render eight seconds of genuinely good motion from a sentence. It still can't cut your ad. Here's the honest line between the two, drawn from four real reels.
Lending money to a friend goes wrong quietly. Here's what actually changes when the debt stops living in one person's memory and becomes a receipt you can both see.
AI can research a prospect and write the email. I won't let it press send. Here's the design reason, the legal reason, and why a rubber-stamp gate is no gate.
A backtest is a description of the past, not evidence about the future. Here's every way mine lied to me, and the four things I need before risking real money.
A business is the worst describer of itself. Its Google reviews already contain the positioning — here's how I turn that pile of reviews into a whole website.
Notes apps and spreadsheets both fail a fragrance collection for the same reason. Here's what tracking one on your phone actually needs to do to be worth it.
Next.js vs FastAPI isn't a performance question when you're building solo. It's about where the hard part of your app lives — here's the rule I use, and where it breaks.
Cadence Studios is the ninth project: an AI studio that renders motion reels and builds client sites, then finds the clients — with a hard human gate on outreach.
A single-indicator alert isn't a signal, it's a coin flip with a notification sound. Here's why they fail and how scoring for confluence fixes each failure.
The awkward part of splitting a bill isn't the math — it's asking twice. Here's how I stopped eating the loss and started keeping a receipt instead.
When building got cheap, the economics of a one-person studio flipped. The cost isn't making things now — it's carrying them. Here's where the weight goes.
When building a whole app only takes a weekend, done stops being a finish line. Here's how I decide something is actually done instead of just running.
claude-blog runs a team of agents that research, structure, write, and optimize a post — built to rank on Google AND get cited by ChatGPT and Perplexity. Here's why I'm handing my blog to it.
The hardest part of vibecoding isn't getting AI to write code — it's verifying it. Here's how I actually test what an AI builds before I trust it.
ClipForge is an autonomous content engine — it turns long video into short clips and posts them across accounts on a schedule, at the volume short-form growth actually takes.
People ask what tools I use like there's a secret stack. There isn't. Here's the honest, boring, short list of what's actually open when I build.
How does a solo developer market seven apps with no marketing team? I'm pointing App Marketing OS — the marketing suite I built — at my own portfolio.
Two updates worth logging: the portfolio is up to eight projects with ClipForge, and the community tool got its final name — CommunityHQ.
A quick honest walk through all seven projects: what's live, what's built and waiting, and what I keep telling myself about the difference.
tab. didn't pass App Store review the first time. Here's what the rejection actually taught me — and it wasn't about code.
Everyone thinks vibecoding means you type a sentence and an app falls out. Seven products later, here's the part nobody tells you.
I can build the apps. Marketing them is the part that quietly kills indie projects. So I built the suite that does the five things I always skip.
Running a Skool community is mostly invisible admin. I automated the grind — and learned a small lesson about names along the way. (It went SkoolOS → CommunityOS → CommunityHQ.)
Everyone hears 'credit' and thinks credit repair. That's not the point. The point is the yes from a lender — and knowing exactly what's standing between you and it.
The fragrance app I'd wanted for years. Point your camera at a bottle, it knows what it is, and the whole thing feels like the shelf it's tracking. (It used to be called Sillage.)
Most signal services fire off a single indicator and call it a setup. Real traders wait for things to agree. So that's what I built.
Every analytics tool tells creators what happened. None of them tell you what to do about it. So I built the one that does.
The first entry. Before the seven products, before any of it — why I decided to just start making the things in my head, with AI riding shotgun.