Inspiration

Marketing orchestration is often fragmented across dozens of tools, creating "cognitive debt" for professionals who spend more time on operations than on strategy. We wanted to build a "Neural Operating System" that acts as an autonomous digital twin, turning complex intent into verified assets instantly.

What it does

AGENTICUM G5 is a high-performance orchestration platform. A user inputs a single campaign goal, and the system triggers a "Synergistic Dialogue" between specialized AI agents. The Orchestrator, Strategist, and Auditor debate the optimal path in real-time, grounded by real-market intelligence, before generating studio-grade visuals (Imagen 3) and professional copy with instant voice synthesis (Neural TTS).

How we built it

The platform is built on a "Neural Substrate" architecture. We used:

  • Gemini 1.5 Pro for the core reasoning and multi-agent synergy.
  • Imagen 3 for high-fidelity visual asset generation.
  • Google Cloud Functions (2nd Gen) for a scalable, high-compute backend.
  • React/Vite/Tailwind for a high-density, low-latency console UI.
  • Google Search Grounding to ensure all strategies are anchored in reality. ## Challenges we ran into Integrating multiple preview-state models (like Veo and Imagen) required sophisticated polling mechanisms and robust error handling. Creating a UI that feels like a real-time Operating System while managing long-running AI operations was a major technical hurdle. ## Accomplishments that we're proud of We achieved a zero-friction workflow where a single prompt results in a full marketing package. The "Neural Dialogue" logs show the AI critiquing its own work, which significantly improves the quality and strategic depth of the final output. ## What's next for AGENTICUM G5 We plan to expand the modular node system to include specialized "Ethics Guardians" and "Market Sentiment Oscillators," and integrate full-length video production once the creative models reach maturity.

Built With

Share this project:

Updates

posted an update

Correction regarding Model Version (Typo)

"Just to clarify for the judges: There is a typo in our project description/tagline mentioning 'Gemini 1.5'.

Please note that our submitted code (as of the Feb 10th deadline) is and has always been running on Gemini 2.0 Flash (Thinking Mode).

This is purely a labeling error in the description text. The actual implementation in the repository ( vertex-ai-client.ts ) correctly utilizes the gemini-2.0-flash-thinking-exp model as originally submitted. No code changes have been made since the deadline.

Happy reviewing!"

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