Inspiration
As a solo builder, I’ve always been fascinated by how brands create ads that grab attention in seconds. But when I tried to create one myself, I realized how painful the process is. You either spend hours learning design tools or pay freelancers/agencies for content that may or may not perform.
That made me think, what if creating high-quality, scroll-stopping ad creatives could be as easy as typing a prompt?
That's where Adlume started. I wanted to build something that lets anyone, no matter their skill level, create stunning images and video ads in just a few clicks, powered by the latest AI models.
Promo Code - BOLTJUDGES100
Use this promo code to receive 100 credits for testing the product.
What it does
Adlume is a SaaS platform where users can:
- Generate high-resolution image ads just by describing their product
- Create short video ads with visuals, captions, background music, and voiceovers, automatically
- Use a credit system to manage usage fairly and efficiently
- Browse, download, and manage all their generated creatives from a personal dashboard
It’s designed to be fast, simple, and powerful—whether you're a startup founder, indie maker, or small business owner.
How I built it
This project was built solo during the hackathon.
- Frontend: Next.js 14 (App Router), styled using TailwindCSS + Shadcn UI
- Backend: Prisma with PostgreSQL for user and credit management
- Authentication: Clerk
- Image Generation: OpenAI's new GPT-based image model
- Video Generation: Pixelverse v4 model + dynamic captioning
- Voiceovers: ElevenLabs integration
- Background Music: Auto-included for video ads
- Payments: LemonSqueezy credit system integration
Challenges I ran into
- Handling delays and async rendering without breaking UX flow
- Ensuring real-time credit deductions worked reliably
- Working with newer AI APIs that lacked extensive documentation
Accomplishments that I'm proud of
- Shipped a working SaaS product with end-to-end functionality - solo
- Integrated multiple AI systems into one smooth workflow
- Built something that feels like a real tool, not a demo
- Delivered a clean user experience with real-world usability
- Handled everything from design to deployment without templates
What I learned
- How to orchestrate multiple AI services into a cohesive product
- Real-world SaaS building with authentication, credits, and billing
- Handling media (video, voice, image) efficiently and in sync
- Why good UX matters just as much as the tech under the hood
What's next for Adlume
- Support for model-based mockups (e.g., products held by models)
- Add platform-specific templates for TikTok, Instagram, etc.
- Custom branding options like logos, color palettes, and overlays
- Video editor improvements (text animation, voiceover control)
- Launch a public ad showcase/gallery
- Introduce team collaboration tools for shared workspaces
Built With
- chatgpt
- elevenlabs
- neon.tech
- nextjs
- postgresql
- prisma
- replicate
- shadcn
- stripe

Log in or sign up for Devpost to join the conversation.