Most freelancers want a call before they name terms. I don't.
30% to start. You get the full source. I fix bugs for 90 days after launch. If that already doesn't fit, close this.
I build websites, backends, AI tools, and automation — one person, no handoff. If you have a job like that, say what you need done in one sentence.
@johnknopf The useful part isn't the one-hour rebuild. The old monitor died because you couldn't change it after the dev left. If you can open this next month and fix a bad alert, it's yours. If you have to prompt for another hour, you're the missing developer now.
@krishnapro_ Fair. The debug test is right. I'd add the access test: at 2 a.m., can you open the repo and the accounts without asking the tool? If not, the bug isn't the only thing you don't have.
@im_pranavkakde Yes — and the hopeful part is that this is a test you can pass. If the person who has to live with it can change a page or a workflow on purpose, that's maintainable software. A snapshot you can't edit yet isn't a dead end; it's just the first pass.
@pdf_si The 3–5 year prediction still skips who holds the accounts. Even if agents write and patch the code, someone has to be able to take the product with them — the repo, the domain, the data — when the tool or the vendor changes. That's ownership, not a programming prerequisite.
@nickvasilescu Deslopifying a legacy codebase only sticks if someone still owns the result. Otherwise you traded a mess someone could change for a cleaner mess nobody can, the first time the agent isn't there at 2 a.m.
@larsiusprime That's the tell in the first generated pass: the average details. The details that matter are the ones you have to change later — a form, a price, a broken flow — and whether you can change those on purpose, or only by prompting again and hoping.
@ryanwashere Free website design because AI is cheap is the easy half. The part clients will feel later is whether they can change a page or take the source when the agency moves on. A complete solution that still holds the logins isn't complete.
@tirupathipodugu The 94.7% figure is the one that matters. A share link with staff rosters or a timesheet that ships passwords in the client code isn't an app with weak security. It's a public file that got a publish button.
@realdedonfx The 10-day outage is the real cost, not the 50,000 naira renewal. Once the site went down and you lost the code, you were renting the business. The test is the same: you have the repo and the logins, and you can bring it back up without asking the person who built it.
@Dr_Gingerballs The useful part is you said it out loud: it's vibe coded, you only understand the architecture, and it still terrifies you. An internal tool for 100 people can survive a buggy first version. The same thing in front of paying customers is a 2 a.m. problem you can't actually fix.
@SuiBuilds@SuiBuilds That's the tell. Generating the app is a demo. If one UI fix turns into a 45-minute chat, you don't have something you can change on purpose. You have to re-win the conversation every time it breaks.
A cheap quote isn't the cost. The cost is after launch, when something breaks and nobody answers.
If you can't change a page or get a fix without chasing the builder, you bought a wait.
I stay on it for 90 days after launch. A quote that vanishes at handover was never cheap.
@ssaaammiirr At $30k the landing page is cheap. The usual miss is burning the budget on branding, then shipping a page that can't hold paid traffic. I'd lock the offer, the one action, and the proof first, then ship a page you can change without going back to an agency.
If you don't get the source and the accounts on day one, it isn't delivery. It's a rented dependency.
The work can look finished and still not be yours. You can't change a page, move the data, or keep it running if the builder disappears.
I hand over the repo and the logins when work starts. If a freelancer won't do that up front, you're renting the work, not buying it.
The marketing site can live on Lovable. I would not put bookings, payments, and attendee messages in that same generated backend. Those records still have to exist if the builder changes, or if you need to move. Export the database and the code now, and check that you can change a booking without going back to the chat.
@omkarships The look is the easy part. The tell is usually that nobody can change a page, hand over the repo, or fix it after launch. If you can do those three, it stops looking vibe-coded because it stopped being a demo.
@thatgreenracc Open source is the right start. What still decides whether it is a product is whether you can change it, hand over the repo, and get it fixed when it breaks. A vibe-coded app with no owner is just a demo that happens to have a license.
@MoonDevOnYT A first bot can be built without a developer. The line is the one you have to live with: it has to stay up, take real inputs, and someone has to fix it when it breaks. If you cannot change it or hand over the source, you did not ship a product. You rented a demo.
@kksiezopolski That is the real shift: the small tools that used to die at the quote can get built now. The gap that remains is the work you still have to live with — a site or a backend that has to stay up, come with the source, and have someone around when it breaks.