I vibe-coded my site — then realized Webflow does it all

How I actually built my site

I want to be upfront about something. When people hear "vibe coding," they imagine someone typing "make me a website" and watching magic happen. That's not what I did.

I knew what I wanted. I understood the architecture — how the frontend talks to the backend, what CMS I needed, how routing should work, what the data structure looks like. I've been in tech long enough to know the difference between a Next.js app and a WordPress site, and why it matters.

Claude Code was my tool. I was the architect.

I used it to write the actual code — React components, API routes, styling, the whole thing. But every decision about structure, technology stack, and integrations was mine. The AI didn't choose my CMS. I did. The AI didn't decide how to handle authentication or SEO. I did. Then I pointed Claude Code at each problem, one by one, and it wrote what I described.

Sounds efficient, right? And it was — until deployment day.

I pushed everything to Vercel, and that's when things got fun. Package A needs version 4.2, but package B breaks above 3.8. The build works locally but fails in production. A dependency I never installed directly is suddenly causing errors because something else updated it. I spent an entire evening fixing things that had zero to do with my website's actual content or design.

Then came the stuff I had to set up manually: server-side optimization, security headers, caching rules, rate limiting. Each one a separate research session, a separate configuration, a separate thing to maintain.

The site launched. It works. But let me put some numbers on the process: 11 Claude Code sessions, roughly 3.5 million tokens burned in total. That's about 2.5 million words of back-and-forth with AI — the equivalent of writing 25 books — just to get one website from zero to production. And that doesn't count the hours I spent between sessions researching, debugging, and configuring things Claude couldn't help with.

Most of those tokens didn't go toward building features. They went toward fighting infrastructure — fixing dependency conflicts, setting up security headers, configuring deployment pipelines.

What Webflow gives you without the headache

The first time I opened Webflow after finishing my vibe-coded site, I felt a very specific emotion. I think the word is "pain."

Everything I assembled from fifteen different sources — stitched together with config files, environment variables, and prayers — exists in Webflow as a single, integrated platform. Just... there. Ready to go.

Hosting. CMS. Forms. Responsive design. Animations. SSL. Sitemap generation. Clean semantic HTML. All of it, built in.

But here's the part that really got me. Webflow doesn't just handle basic SEO — the meta tags, the sitemaps, the heading structure. It handles AEO out of the box. Answer Engine Optimization — the stuff that determines whether ChatGPT, Perplexity, or Google's AI Overviews pick up your content. Clean semantic markup, structured data, fast load times, proper content hierarchy. With my vibe-coded site, I had to research each of these things, implement them separately, and hope I didn't miss anything. In Webflow, it's just how the platform works by default.

Let me put it in real terms. Here's what specific tasks look like in both worlds:

Adding a blog post. Vibe coding: open VS Code, create a new markdown file or CMS entry, make sure the routing picks it up, commit, push, wait for Vercel to build, check if it deployed correctly. Webflow: open the Editor, write, hit publish. Done in the time it takes to make coffee.

Changing a button color. Vibe coding: find the component, update the style, commit, push, deploy. Or if you're unlucky and it's in a design token system — update the token, rebuild, redeploy. Webflow: click the button, pick a color, publish. Three seconds.

Making the site responsive. Vibe coding: write media queries, test in Chrome DevTools, pray it works on real devices, discover it doesn't, fix, redeploy. Webflow: switch between breakpoints visually, drag things around, see changes in real time.

"The real cost of vibe coding isn't the AI tokens — it's the hundred small decisions you have to make about things that aren't your product."

I'll be fair about the limits, though. If you need complex custom logic — user authentication flows, real-time data processing, integrations with niche APIs — Webflow will hit a wall. It's not a full-stack application platform. But for content-driven sites, marketing pages, portfolios, blogs? It's not even a close comparison.

The long game: why Webflow wins for real projects

Here's the thing I didn't understand until I lived both sides: vibe coding is a prototyping tool. Webflow is a business tool. They solve different problems at different stages.

When you're testing an idea — when you don't even know if anyone will care — vibe coding is perfect. Zero cost, maximum speed, throw it out there and see what happens. That's exactly what I did with my site. It's a test. I'm checking if this concept has legs before I invest anything serious.

But the moment a project moves past the "let's see if this works" phase — when real traffic shows up, when you need lead forms that actually work, when content needs to be updated regularly — Webflow starts playing a completely different game.

Non-technical people can use it. And this is way more important than most developers realize. When your content person needs to publish a blog post, they shouldn't need to learn Git. When a client wants to update their phone number, that shouldn't require a developer. In Webflow, they open the Editor, make the change, and publish. With a vibe-coded site? That's a Slack message, a code change, a review, a deploy — and a two-day turnaround on a ten-second edit.

Real collaboration. Designers, copywriters, SEO specialists, marketing people — they can all work on the same Webflow project without breaking each other's work. Try onboarding a marketing person to a Next.js repo. You'll spend more time explaining what npm install means than actually getting work done.

It scales with the business. Need a new landing page for a campaign? Drag, drop, publish — maybe thirty minutes. Need to A/B test a headline? Just change it and track. Need to add a booking system or a contact form? Built in. Every new feature in Webflow is a visual operation. Every new feature in a coded project is a development cycle with its own dependency tree and deployment risk.

Vibe coding can do all of this too. If you understand the architecture, if you're patient, and if you're willing to burn the tokens and the time. But it takes significantly longer, and you're the single point of failure for everything. Webflow removes about 80% of that infrastructure pain. You just build what you see, and then you ship it.

The honest math: why I'm still on Vercel

Okay. After everything I just said — why am I not on Webflow right now?

Let me lay out the numbers.

My current setup costs $20/month. That's my Claude Pro subscription — the tool I used to build the site. Over the course of development, I ran 11 sessions averaging around 300–350 thousand tokens each. That's roughly 3.5 million tokens total — all covered by the Pro plan, but still: 3.5 million tokens is a lot of work for something Webflow ships out of the box.

Vercel hosting? Free. Custom domain? Free. Analytics? Free. The only recurring cost is the $20/month Pro plan that I'd be paying anyway for other work.

What would Webflow cost me? After the May 2026 pricing update, the Premium plan runs $25/month billed yearly, or $39/month if I pay month-to-month. That gets me everything — hosting, CMS with 20,000 items, forms, 50 GB bandwidth, built-in SEO and AEO, and the full visual editor. No terminal. No build system. No dependency tree.

So the real comparison isn't $0 vs. $25. It's $20/month plus hours of maintenance vs. $25/month and zero maintenance. When I frame it that way, Webflow is actually cheaper. I'm just paying in time instead of money right now — and pretending that's free.

But here's my situation: my project is still in test mode. I'm validating whether this idea works before I scale it. I don't want to commit to monthly infrastructure costs until I see real traction. When you're bootstrapping, every subscription is a bet — and I want to place my bets on things that are proven, not promising.

The trade-offs of the free path are real, though. Security is my responsibility. Performance optimization is manual. If something breaks at 2 AM — I'm the support team. These are costs too, just hidden ones.

"The question isn't which tool is better. It's which tool matches the stage you're at."

The moment I see consistent traffic, real engagement, and signs that this project has legs — I'm moving to Webflow. Not because vibe coding failed. It didn't. But because maintaining infrastructure isn't what I want to spend my time on. I want to write, create content, and grow. Webflow lets me do that. A custom-coded site wants me to also be a DevOps engineer.

Written by Vlad UsenkoGrowth marketer who codes. I write about Webflow, marketing strategy, and building a one-person business.