BrandPrint-ai marketing full automation from YOUR bussiness

The first autonomous creative core powered by Amazon Nova to forge ready-to-post viral videos and SEO packages from any digital seed.

Comment

💡 Inspiration Marketing for small businesses and startups is often a massive bottleneck. We wanted to build an "Autonomous Creative Director" that doesn't just suggest ideas but actually sees a brand, thinks about the strategy, and acts by producing high-fidelity video content and SEO clusters without human intervention. The launch of Amazon Nova provided the perfect multimodal stack to turn this vision into reality—allowing us to move from static tools to a living "Creative Core."

⚙️ What it does AdGen AI is an end-to-end marketing synthesis engine that replaces a full creative team:

Autonomous Scanning: Uses Amazon Nova Act to navigate and "shred" any website, capturing its visual DNA and structural layout. Multimodal Analysis: Amazon Nova Pro analyzes the captured frames to extract branding, HEX codes, audience personas, and unique value propositions. Viral Forging: Amazon Nova Lite generates high-velocity viral video scripts, 600-word SEO blog posts, and platform-specific social media captions. Autonomous Production: Our custom high-fidelity production engine uses Amazon Polly (Neural Voice) and FFmpeg to render real .mp4 Reels/TikToks. These include Ken Burns animations, cinematic overlays, and synchronized AI voiceovers. Neural Native Dashboard: A futuristic "Iron Man-style" interface with biometric animations that allow users to preview generated reels inside a virtual phone and "Post to Social" with a single click. 🛠️ How we built it Frontend: React 18, Vite, and Vanilla CSS with a "Neuromorphic-Dark" aesthetic. Backend: Node.js (Express) orchestrating a multi-stage AI pipeline. AI Core (Amazon Bedrock): Nova Pro: For visual brand reasoning and multimodal analysis. Nova Lite: For high-speed text/script synthesis. Nova Act: For agentic website exploration. Titan Image Generator: For cinematic high-end logo generation. Voice Engine: Amazon Polly (Neural Engine) for professional-grade voiceovers. Media Pipeline: A custom node-canvas and fluent-ffmpeg pipeline for real-time MP4 rendering. Cloud Stack: AWS S3 for video storage, Firebase for real-time data sync, and ImageKit for optimized asset delivery. 🚧 Challenges we ran into Multimodal Synchronization: The hardest part was aligning the visual "Key Visuals" extracted by Nova Pro with the timing of the audio scripts generated by Nova Lite and spoken by Polly. We built a custom "Tease Logic" to ensure subtitles and images shifted in sync with the AI's narrative flow. Server-Side Video High-Fidelity: Rendering 1080x1920 vertical video with cinematic filters and scan-line effects on a server budget required heavy optimization of our Canvas-to-FFmpeg stream. Agentic Navigation: Fine-tuning Nova Act to ignore cookie popups and actually find the "meat" of a landing page required iterative prompt engineering for the navigator agent. 🏆 Accomplishments that we're proud of Zero-Click Production: We achieved a flow where a user simply enters a URL and receives three ready-to-post high-fidelity MP4 videos in under 3 minutes. Holographic UI Aesthetics: The "Biometric-Deep-Scan" login and the "Content Engine" phone mockups look and feel like state-of-the-art software. Deep Nova Integration: We didn't just use Nova as a "wrapper"; we used the specific strengths of Pro, Lite, and Act in a chained agentic workflow. 📚 What we learned Foundation models like Amazon Nova Pro allow for "Visual Reasoning" that previously would have required thousands of lines of manual scraping rules. "Agentic Creative" is viable. Moving from "generative text" to "ready-to-post media" is the key to providing real value to businesses. The unified API of Amazon Bedrock makes switching between Lite (speed) and Pro (logic) extremely frictionless during development. 🔮 What's next for AdGen AI Autonomous Social Publishing: Integrating Direct Publishing APIs for TikTok and Instagram to allow for a fully hands-off marketing department. A/B Feedback Loop: Using engagement data (likes/views) to feed back into Nova Lite to "self-correct" and improve the next batch of generated scripts. Predictive Scheduling: Determining the best post times using Nova's analysis of a brand's specific industry trends.

Built With

Share this project:

Updates