FenixBlack.ai: AI Marketing Agency Platform
7 production apps in 4 months—platform engineering at speed
The Challenge
AI marketing tools in 2024 are fragmented—one tool for copywriting, another for images, another for video. Each with different UI, pricing, and quality levels.
What marketing agencies actually need:
- Unified platform for all creative AI tasks
- Consistent quality across different AI models
- Fast iteration on client work
- Scalable deployment for multiple clients
The approach: Build a modular platform that can deploy specialized apps quickly.
What I Built
FenixBlack.ai - AI marketing agency platform with 7 production applications shipped in 4 months (2024).
Main Platform (Python + NiceGUI):
- Interactive video avatars for client presentations
- AI brand designer for complete brand identities
- Research agents for market analysis
- Custom React bridge - novel integration allowing React components in Python apps
6 Specialized Micro-Apps (Next.js):
- brand.fenixblack.ai - Complete brand kit generation
- backgrounds.fenixblack.ai - Zoom/Teams background creator
- holidays.fenixblack.ai - Animated holiday video generator
- restore.fenixblack.ai - AI photo restoration & animation
- canvas.fenixblack.ai - Hand-drawn animation generator
- growth.fenixblack.ai - Credit and usage management
Live: fenixblack.ai↗
Technical Innovation
The hardest problem wasn't integrating AI models—it was building a Python backend that could render modern React UIs.
The challenge: NiceGUI (Python UI framework) is great for AI prototyping but limited for complex interactions. React has the components I needed but doesn't integrate with Python backends naturally.
The solution: Built a custom React-Python bridge that:
- Renders React components inside Python applications
- Maintains Python's rapid AI integration capabilities
- Keeps React's modern UI/UX patterns
- Enables real-time bidirectional communication
This isn't a library—it's a custom architecture I designed specifically for this use case.
Result: Rapid AI workflow development with production-quality UI.
Multi-App Strategy
Instead of one monolithic application, built 6 independent micro-apps on Vercel.
Why this works:
- Faster iteration - Update one app without touching others
- Specialized marketing - Each app has its own landing page and SEO
- Independent scaling - High-traffic apps don't impact others
- Easier pricing - Customers can buy just what they need
Stack per micro-app:
- Next.js 14 for performance and SEO
- Vercel for instant global deployment
- Different AI models per use case (OpenAI, Stable Diffusion, FFmpeg)
- Shared authentication and credit system
Development Timeline
Month 1: Main platform architecture + React bridge Month 2: First 3 micro-apps (brand, backgrounds, holidays) Month 3: Remaining 3 micro-apps (restore, canvas, growth) Month 4: Integration, testing, production deployment
How I shipped 7 apps in 4 months:
- Shared component library across all Next.js apps
- Standardized deployment pipeline (Vercel)
- Reusable AI integration patterns
- Focused on MVP features first, polish later
Key insight: Micro-apps let you ship incrementally. Each app was usable on day one of its development.
Production Metrics
Current status:
- All 7 apps in production
- Multiple AI models integrated (OpenAI, Stable Diffusion, etc.)
- Custom React-Python bridge working in production
- Independent deployment and scaling per app
What this proves:
- Multi-app platforms can ship fast
- Python + React can work together elegantly
- Complex AI integrations can be standardized
- Micro-services architecture works for AI products
What I Learned
On polyglot development: Don't fight your tools. Python is best for AI integration, React is best for modern UI. Building a bridge between them was faster than fighting either's limitations.
On multi-app platforms: Splitting into micro-apps was the right call. Development complexity goes up slightly, but deployment flexibility and iteration speed go up dramatically.
On AI model integration: Every AI model has quirks. OpenAI for text, Stable Diffusion for images, FFmpeg for video—use the right tool for each job, then abstract the complexity away from users.
Build With Me
Building multi-product AI platforms or need custom architecture? I specialize in:
- Rapid prototyping with multiple tech stacks
- Novel integration patterns (like the React-Python bridge)
- Multi-app deployment strategies
- Production AI systems
Let's talk about your next platform.