I Built My Website With AI. Do I Need a Developer?

TL;DR. Many AI-built sites are fine. Brochure site, low traffic, no payments, and you're happy tinkering: keep it, spend nothing. Hire when the site touches revenue, forms fail silently, it needs to convert, or it's actually an app with logins and payments. Paste the decision prompt below into your AI, answer up to seven questions, and get a plain verdict: stay DIY, hire for a fix, or rebuild. Rebuilds start at $399. App cases start with the $500 Week-1 Build Audit.

The honest answer: many AI-built sites are fine

Here's the case for keeping it, since nobody selling web design will make it for you. If your site is a brochure, who you are, what you do, how to call you, and it gets modest traffic, takes no payments, and you don't mind poking at it when something needs changing: keep it. Spend nothing. The AI builders are genuinely good at that job now, and a developer would mostly be charging you to feel safe.

The hire cases are just as specific. The site touches revenue: bookings, quotes, sales, leads you'd actually miss. Forms that fail silently, the classic being a form that says thanks while the email goes nowhere, and you find out a month of enquiries later. A site that needs to convert, not just exist. Or the big one: what you built isn't a website at all, it's an app with logins, payments, and other people's data. That last case changes everything, and it gets its own section below.

Buyers already route themselves this way. A recent DappaSol lead found us by asking Gemini for help with their problem, not by Googling agencies. Ask your AI the right way and it gives you a straight answer. The prompts below are the right way.

The decision prompt: paste this into ChatGPT, Claude, or Gemini

Copy this into a fresh chat. It asks up to seven questions, one at a time, and it's built so that staying DIY is a real possible outcome, not a polite pause before the upsell.

You are a senior web engineer who has reviewed hundreds of AI-built sites, and you have no stake in whether I hire anyone. Interview me about the website I built with an AI tool, then tell me straight whether to keep it, fix it, or rebuild it.

Rules for the interview:
- Ask ONE question at a time and wait for my answer. No multi-part questions.
- Ask at most 7 questions. Skip any you can already answer from what I've said.
- Push back once when an answer is vague, then move on.

Cover, in this order: what the site is for (brochure, leads, sales, bookings); traffic today and where I want it in 12 months; how much revenue or how many leads run through it; who maintains it and how often it needs changes; what's broken or missing right now; my growth plans for the next year; how comfortable I am fixing things myself.

Then output a decision sheet with exactly these sections:
1. Verdict on the first line: STAY DIY, HIRE FOR A FIX, or REBUILD, with one sentence of why
2. Decision matrix: one row per factor (purpose, traffic, money at stake, maintenance, breakage, growth, technical comfort), my answer in five words or fewer, and whether that factor points DIY or hire
3. If STAY DIY: the three things I should fix myself, in priority order
4. If HIRE FOR A FIX: the exact job, in one sentence I can send to a developer
5. If REBUILD: the three requirements the new site must meet that this one can't
6. What would change your verdict, in one line

Keep the whole sheet under 300 words. Recommend hiring only where my answers justify it. STAY DIY and spend nothing is a fully acceptable outcome.

Copy this prompt

Answer honestly, especially the money question. The verdict swings almost entirely on how much revenue runs through the site and whether anyone would notice if it broke on a Tuesday. If your answers are none and no, expect to be told to keep your money.

What each verdict means

Stay DIY. The matrix pointed at a brochure site with no money on the line. Fix the three things the sheet listed, keep your login details somewhere safe, and get back to running the business. Revisit only when the site starts to matter to revenue. Total cost: nothing, and that's the correct amount.

Hire for a fix. This is a scoped job, not a relationship. Take the one-sentence job description the prompt wrote and send it to two or three developers for a fixed quote. Hire someone to build a website maps every route and what fair looks like, and small business website cost has the price bands so nobody rounds you up. If a quote comes back hourly with no cap, that's a running meter. Walk.

Rebuild. Usually the verdict when the site needs to convert, the builder is fighting you, or the fix list keeps growing back. Don't feel bad about it. You validated the business with a cheap tool, which is more than most people do before spending real money. The rebuild routes and exact prices are at the bottom of this page.

The reality-check prompt: assume it breaks

Maybe the interview said stay DIY and you're ready to trust the site. Don't trust it. Verify it. AI-built sites fail quietly: the demo looks perfect, and the failure modes, a form posting to nowhere, a site nobody backs up, pages Google can't read, only surface when they cost you something. This prompt audits like the person who gets blamed.

I built my website with an AI tool: [name the tool and paste your URL]. Audit it like the engineer who gets blamed when it fails.

Assume the site fails under real traffic, real SEO scrutiny, and a real security check until my answers prove otherwise, and say SOLID only on specific evidence, never on my enthusiasm.

Ask me for the evidence you need, one item at a time: what happened when I last submitted my own contact form to an inbox I check, whether a backup exists outside the builder and has ever been restored, how the site loads on a phone on mobile data, where my analytics live and what they show, and anything else you need to score honestly.

Then output a risk report with exactly these six checks, each scored 1 to 5 with one line of evidence:
1. Mobile performance: loads fast on a real phone on mobile data
2. Forms: a test submission actually arrived, verified this month
3. Backups: one exists outside the builder and a restore has been tested
4. Analytics: I can see visitors, sources, and conversions
5. SEO structure: real page titles, one H1 per page, indexable, sitemap submitted
6. Legal pages: privacy policy, terms, and cookie handling that match what the site does

Score only what I gave evidence for and mark everything else UNVERIFIED, never average. End with one overall verdict: SOLID, FIX THESE FIRST with the list in priority order, or AT RISK, and name the single check that worries you most.

Copy this prompt

The forms check is the one that catches people. Submit your own form to an address you actually read, today, before you do anything else on this page. And when you reach the SEO check, run the sibling test too: does ChatGPT recommend your business shows what the AI engines say when a customer asks for someone like you. Weak structure on the site is usually why the answer is nothing.

Broken already? Sort the bugs before you pay anyone

Typed "my ChatGPT built app is broken" into a search bar at 11pm? This section is for you. Don't hire in a panic, and don't paste forty bugs into an agency contact form, because the quote that comes back will be priced on your desperation. Sort first.

My AI-built site has problems. Here's the list: [paste every bug, broken feature, or thing that looks wrong, one per line]

Sort every item into exactly one of three buckets. Never default the whole list to "hire someone": if something is a ten-minute fix, say so and tell me how.

Output a table with these columns: Item, Bucket, What to do, Effort.
- QUICK DIY FIX: name the exact fix and where to click or what to change. Assume I follow instructions well but don't write code.
- NEEDS A DEVELOPER: describe the job in one sentence I can paste into a message to a developer, and give a fair one-off price band for that job at freelance market rates.
- REBUILD TERRITORY: use this when fixing it properly means changing how the site was built, not patching what's visible.

After the table: count the buckets. If three or more items are REBUILD TERRITORY, or the developer price bands add up to more than a simple professional rebuild, say plainly that patching is the expensive option. Finish with the one item I should handle today and why that one.

Copy this prompt

Anchor the developer price bands against a real number before you accept any of them: a full professional rebuild is $399 with us, and the Engine starts at $699 for a site plus a working lead pipeline. If patching the old site quotes higher than replacing it, you have your answer. And if what's broken involves logins, payments, or customer data, stop patching and read the next section, because that's not a website problem.

When your "website" is actually an app

If people log in, pay, or store their data in the thing you built, you didn't build a website. You built an app, and the DIY-versus-hire question changes shape completely. Websites fail visibly and cheaply. Apps fail quietly and expensively: a leaked customer table, a payment that half-completes, a database the AI tool left open to the internet.

Don't route that through a web designer, ours included. Start with is my AI app production-ready for the honest checklist, then take your AI app to production for the fix path. If you're still scoping what the thing even is, the app requirements prompt turns your idea into a one-page sheet a developer can price.

When you want a human verdict, that's the $500 Week-1 Build Audit: one week, a one-page spec, a fixed price, and an honest go or no-go, credited in full if we build. If the audit says your prototype is fine as it stands, it says so. We've shipped 100+ products, with team history at ShapeShift, CoinDesk, and WallStreetBets, and the audit exists because "just rebuild it" is sometimes the wrong answer too.

If the verdict is rebuild

Two fixed prices, no proposal theatre. The Overnight Site is $399 and live in a day; website in a day shows exactly how that works without cut corners. The Engine starts at $699 and takes 2 to 4 days: the site plus a lead pipeline, so forms land in a CRM, bookings get captured, and follow-up happens without you remembering to do it. That's the cure for the "my form said thanks and nothing arrived" era of your site.

Either way, you own 100% of the code from day one, the price is fixed before we start, and there's a 30-day warranty. No builder subscription, no platform holding your site hostage, nothing that breaks because a template updated underneath you.

One thing before anyone builds, us or not: spend ten minutes on the brief. The plan-a-website prompt interviews you the same way the decision prompt did and hands you a one-page brief any builder can quote from. A clear brief is the cheapest upgrade a rebuild can get.