Brenda Jazmín: From Influencer to Owned Course Platform

"No quiero pagar Kajabi por siempre. Quiero que sea mío."

Client: Brenda Jazmín (@brendaa_jazmin) — Mexican fitness influencer Goal: Replace Kajabi with a fully-owned course platform for "Cake Pump Workout" Stack: React + Vite + Supabase + Stripe + Claude API + Vercel serverless proxy

The problem

Brenda built an engaged audience in the Mexican fitness niche and was ready to sell her signature program. Like most creators, her options were:

• Kajabi/Thinkific: Easy to launch, but ~$150–300/month in perpetuity plus transaction fees on some tiers
• Stan Store / Linktree: Inexpensive but limited customization and basic UX
• Custom build: Higher upfront investment, owned forever

She chose owned. We built.

What we built

Page 1 · Landing / Sales page

Cinematic hero with brand video. Hook copy written in Mexican Spanish — Brenda's voice, not translated. Social proof carousel with transformation photos. Curriculum preview. Bilingual testimonials. Sticky CTA on mobile. Stripe-powered checkout integration.

Page 2 · Checkout

Direct Stripe integration — no third-party processor middleman. One-page checkout. Auto-creates user account on successful payment. Email confirmation in Spanish. Tax handling for MX/US buyers.

Page 3 · Course content platform

Video player with progress tracking (resumes where you left off). Module/lesson structure with marked-complete states. Downloadable materials (PDFs, calendars). Notes section per lesson. Bilingual UI. Mobile-optimized — most fitness customers watch on phones during workouts.

Bonus: "Brenda AI Vision" companion app

Separate app with 8-step onboarding flow (goals, current fitness state, preferences). Personalized workout plans generated via Claude API. Personalized meal plans, also AI-generated and bilingual. Progress check-ins.

Critical architecture note: Claude API integration requires a serverless proxy — direct browser → Anthropic API calls fail due to CORS. We built the proxy as a Vercel serverless function. Now standard in our internal playbook for all future AI integrations.

Vercel Root Directory configuration

Vercel's Root Directory setting caused the initial deploy to fail due to the monorepo structure. Fix: explicit Root Directory in Vercel project settings + vercel.json with correct output config. Now standard in all our deploys.

Results

Ownership:

• Fully owned platform — no monthly SaaS fees to Kajabi or Thinkific
• Complete customer data ownership (emails, behavior, completion rates)
• Branded experience matching Brenda's aesthetic, not a template's

Technical delivery:

• Launch-ready in 6 weeks
• Bilingual native UX — Spanish-first, not translated
• AI Vision companion app as post-purchase upsell, increasing customer LTV
• Vercel serverless proxy pattern documented for reuse across future AI projects

Creator economics:

• One-time investment replaces indefinite SaaS subscription
• Direct Stripe integration eliminates middleman transaction fees
• Customer data stays with Brenda, not a platform

What we learned

Own vs. rent math depends on audience size. For creators with a large, engaged following like Brenda, the custom build pays back within months. For smaller audiences, off-the-shelf tools still make sense. We tell clients honestly which category they're in.

AI API integration needs a server layer. The CORS limitation isn't obvious until you hit it. Building the serverless proxy added ~2 hours but unblocked the Claude API integration entirely. Now it's day-one infrastructure on every AI project.

Bilingual UX means Spanish-first, not bilingual-second. Writing the Spanish copy first, then adapting to English, produces a fundamentally different result than translating. The voice stays authentic.

Tech stack

MORE CASE STUDIES