Ashikur Rahman
SEO & AI Workflow Specialist | Claude Code Expert
New to Contra
Ashikur is ready for their next project!
I built my entire portfolio site with 10 Lovable credits. Here's the exact workflow.
Most people burn through Lovable credits fast because they try to build the whole thing inside Lovable, prompt after prompt after prompt. I don't. Here's how I kept it to 10.
The workflow: Lovable + Claude Code + GitHub
1. Lovable for the scaffold only. One clean generation to get the project skeleton, routing, and a starting design. That is where the ~10 credits went, and that is the last time I spent any.
2. Claude Code does the real building. Every page, the whole design system, the content, the SEO, the performance work, all of it happens in Claude Code on my machine. No per-prompt credit meter running.
3. GitHub is the bridge. Claude Code commits and pushes straight to the repo.
4. Lovable auto-syncs from GitHub. The changes land back in Lovable with zero extra credits burned.
So Lovable stays the front door and the deploy target, but the heavy lifting lives in Claude Code, where I am not paying by the prompt.
What that actually produced (in the video):
- A 15-page site where every page is an original design, not a template reskin
- 100 PageSpeed on desktop, every category green
- Built for AI search, not just Google. llms.txt, clean structured data, the full AEO setup
- Self-hosted fonts, accessibility at 100, real Core Web Vitals work
Live at hey-ash.com (http://hey-ash.com).
Loveable Link: hey-ash.lovable.app
If you are spending real money on Lovable credits, this is the unlock. Scaffold in Lovable, build in Claude Code, sync through GitHub. Happy to break down any part of it.
0
36
Most builds here start in Figma or Framer. This one started in a terminal.
I am an SEO specialist, not a designer by trade, but I built and shipped my whole agency site, Proximity Ranker, entirely in Claude Code. No page builder, no template. Just prompts, a clean blue design system, a sticky pill nav, and a proof section that pulls real Google Search Console and Bing data instead of stock testimonials.
The whole site is built to rank on Google and get cited by AI, which felt like the honest way to sell what we actually do.
Quick scroll in the clip. Would love your thoughts on the build.
1
1
96
1.86M Search Impressions and 10.8K Clicks in 3 Months for a Dermatology Clinic
Google Search Console showing 1.86M impressions and 10.8K organic clicks over 3 months at an average position of 9.4. Driven by technical SEO, content optimization and AEO, with workflows run through Claude Code.
0
39
1,083 Calls Driven from Google Business Profile for a Dental Clinic
Local SEO that produces real customers: 1,083 calls made directly from a client's Google Business Profile, up 15% year over year. Result of Google Business Profile optimization, local schema and review strategy.
0
44
Headless React Site on Cloudflare: Built with Claude Code
Rebuilt a content site off WordPress into a headless React/Astro stack, deployed on Cloudflare with a Git → GitHub → Cloudflare auto-deploy pipeline. Editorial design, fast Core Web Vitals, and a fully automated publish workflow; designed and built end-to-end with Claude Code.
0
54
SEO Agency Homepage: Designed & Built with Claude Code
Conversion-focused homepage design for a local SEO and AI-search agency targeting doctors, lawyers, and local businesses. Bold editorial typography, trust signals (reviews, proof stats), clear CTA hierarchy, and a clean blue/navy palette. Designed and built end-to-end with Claude Code.
0
52
SEO Growth on Autopilot: 190K Impressions via Claude Code
Google Search Console showing organic growth from near-zero to 190K impressions in 3 months. Content production, technical SEO, and publishing automated end-to-end with Claude Code; programmatic page builds, schema markup, and AEO-optimized articles shipped at scale.
0
56
189.9K AI Citations Across ChatGPT & Copilot
Bing Webmaster AI Performance report showing 189.9K citations across Microsoft Copilot, ChatGPT search, and partner AI engines over 3 months, with an average of 118 pages cited. Achieved through answer-engine optimization (AEO): answer-first content structure, entity and schema markup, and extractable facts that AI assistants pull as sources.
0
75