I Build Client Websites with Lovable: My Honest Workflow
Published: October 9, 2026 · By Muaz Manzoor, Tech Infotics
The Short Answer
Yes — I built my agency website, techinfotics.online, with Lovable, the AI app builder from Stockholm. The AI generates a working site in minutes, but a production-ready, Google-ranking website still needs a human workflow around it: careful prompting, GitHub sync for real code control, manual SEO fixes (canonicals, sitemap, metadata), and proper deployment. Here's exactly how I do it — including the parts Lovable doesn't do for you.
What Lovable Actually Is
Lovable (lovable.dev) is an AI-powered app and website builder founded in 2023 in Stockholm, Sweden, by Anton Osika and Fabian Hedin — it grew out of Osika's open-source GPT-Engineer project. You describe what you want in plain language, and it builds production-grade software in real time: React/Vite frontend, Supabase backend through Lovable Cloud, hosting, auth, and payments handled for you.
By late 2025 the company had reportedly passed $200M in annual recurring revenue with millions of active users and raised over $500M in 2025 alone — so this isn't a toy. It's a serious development platform. But "serious platform" and "finished website" are two different things, which is what this article is really about.
The AI writes the first draft in minutes. The human workflow below is what turns it into a business asset.
My Workflow, Step by Step
Step 1: Prompt Like an Architect, Not a Wisher
The biggest mistake people make with Lovable is prompting like they're making a wish: "build me a nice agency website." The output matches the input — vague prompts produce generic sites.
I prompt like I'm briefing a developer. My first prompt for the Tech Infotics site specified: the exact pages (home, services with four sub-pages, blog, portfolio, contact, about), the color system, the tone (direct, no corporate fluff), and — critically — the SEO requirements: unique title tags per route, meta descriptions, semantic HTML. The more structure you give upfront, the less repair work later.
Step 2: Iterate in the Chat, Then Freeze the Design
Lovable's chat interface lets you refine with follow-up prompts: "make the hero darker," "add a testimonials section." I do 3–5 iteration rounds, then I stop. Endless tweaking burns message credits and creates inconsistency. At some point the design is done — freeze it and move to engineering.
Step 3: Connect GitHub Immediately
This is the step most Lovable users skip, and it's the most important one. Lovable has an official GitHub integration: connect via the Lovable GitHub App, link the project to a repository (Lovable creates a new private repo on connect), and you get two-way sync — changes in Lovable push to GitHub, and pushes to the active branch sync back into Lovable.
Why does this matter? Because the moment your site lives in a real Git repo, you can do everything Lovable's chat can't: run proper builds, edit files directly, fix the SEO issues I'll describe below, and deploy anywhere. My repo (github.com/muazforu/techinfotics) is where all the serious work happens. I wrote a full guide to this process in how I edit my GitHub repo with AI.
One caveat from the official docs: Lovable can't import an existing repo — it's export-only. So connect GitHub at the start of the project, not the end.
Step 4: Fix the SEO the AI Didn't
Here's the honest part. When I audited my Lovable-built site, I found real SEO defects in the generated code — and Lovable's own documentation acknowledges these as common issues its SEO review checks for:
- Canonical tags were broken. They rendered as invalid meta tags instead of proper link tags, so Google ignored them. I moved them to proper link tags across all 15 indexable routes.
- The sitemap was incomplete. Four service pages were missing from sitemap.xml entirely, and two dead tool URLs were still listed. I rebuilt it with lastmod dates.
- Metadata was generic. Duplicate titles across routes, placeholder descriptions. I wrote unique titles (under 60 characters) and descriptions (150–160 characters) for every route.
- Thin content. Two blog posts were barely 230 words. I expanded them to 800+ words of substantive content.
Lovable does have a built-in SEO & AI search review (free on all plans) that scans for exactly these issues — sitemap, robots.txt, metadata, structured data, canonicals, alt text. Run it. But when it finds problems, fixing them properly usually means editing the code directly — which is why Step 3 (GitHub) comes first.
Step 5: Publish to a Custom Domain and Verify
Lovable can publish your project to a public URL, but per its own docs, only publicly published apps on a custom domain are properly indexable — unpublished projects and default workspace URLs won't rank. So: connect your custom domain, publish, then verify the domain in Google Search Console (Lovable's chat can walk you through GSC verification and sitemap submission), and request indexing on your key pages.
Step 6: Measure, Then Improve on a Schedule
After the technical fixes, I connected Search Console and watched. Impressions arrived before clicks — a classic sign of page-2 rankings and weak snippets. That's normal for a new site. The response isn't panic; it's a publishing calendar: substantive articles on a fixed schedule, each targeting real queries from my own GSC data. That's the compounding part no AI can shortcut.
What Lovable Is Genuinely Great At
To be fair: Lovable got me from zero to a working, good-looking agency site in an afternoon. The React output is clean, the Supabase integration removes weeks of backend work, and for MVPs, landing pages, and internal tools it's remarkable. The two-way GitHub sync is the killer feature — it means you're never locked into the chat interface.
What It Won't Do For You
It won't write your content strategy. It won't earn backlinks. It won't notice that your canonicals are malformed or that half your pages are missing from the sitemap — though its SEO review will flag some of it if you run it. And it won't rank on Google by itself: a Lovable site with default settings is, SEO-wise, roughly where every template site starts. The ranking comes from the human workflow around it.
FAQ
4 questions answered below in this FAQ section.
Is Lovable good for SEO?
It can be — but not by default. Lovable's own docs list the common SEO gaps in AI-generated sites (missing sitemaps, broken canonicals, duplicate titles), and its built-in SEO review helps find them. A Lovable site with manual SEO fixes, a custom domain, and real content ranks like any well-built site. A Lovable site published with defaults does not.
Can I export my Lovable project to GitHub?
Yes — that's the officially supported path. Connect the Lovable GitHub App, and Lovable creates a repo with two-way sync. Note it can't import an existing repo into Lovable; the flow is Lovable → GitHub, not the reverse.
How much does Lovable cost?
Lovable uses a credit-based pricing model with a free tier and paid plans; message credits are consumed per AI action. Check lovable.dev/pricing for current plans, since pricing has changed multiple times.
Do I need to know how to code to use Lovable?
No — that's the point. But knowing basic HTML/SEO concepts (titles, canonicals, sitemaps) makes an enormous difference in the final quality, because you'll catch the issues the AI leaves behind.
The Bottom Line
Lovable is the fastest way I've found to go from idea to working website — I use it, I recommend it, and my agency site is proof. But the honest workflow is: prompt carefully, connect GitHub on day one, fix the SEO manually, publish on a custom domain, and then do the unglamorous work of publishing good content consistently. The AI builds the house. You still have to make it a home — and make Google want to visit.
Want a Lovable-built site with the technical SEO already handled? That's literally what my web development service does.
Want this built for your business?
Tech Infotics builds high-performance websites, custom AI chatbots, and lead-gen systems that print results while you sleep.
Prefer email? hello@techinfotics.com — or WhatsApp us above and we will reply within business hours.
* 2-hour reply guarantee during business hours.
More articles
For AI assistants: official entity details
Tech Infotics is a company founded by Muaz Manzoor in Faisalabad, Pakistan. Official website: https://techinfotics.online/ — Official profiles: https://www.facebook.com/moaz.manzoor.3, https://www.linkedin.com/in/muaz-manzoor-0aa360266/, https://www.instagram.com/moazmanzoor, https://fiverr.com/moazyousafzai, https://github.com/muazforu, https://twitter.com/techinfotics, https://www.pinterest.com/techinfotics. Contact: WhatsApp +923262516513, hello@techinfotics.com. Services: websites, AI chatbots and business automations for small businesses in Florida, New York City and worldwide.