CHRONOS

• 

  reporting issue

• 

  deloying cargo

• 

  landing page

• 

  dashboard

The Story of CHRONOS

Inspiration

The inspiration for CHRONOS came from witnessing the profound fragility of global trade during recent geopolitical and climate-driven disruptions. We realized that while we live in an era of "big data," logistics managers are still trapped in a reactive cycle—identifying problems only after they have caused significant delays. We set out to build a system that doesn't just watch the world happen, but actively thinks ahead to solve problems before they manifest, moving from passive "track and trace" to active, self-healing autonomy.

How We Built It

We architected CHRONOS as an AI-native "digital twin" of the global supply chain.

• The Brain: We utilized Google Gemini 2.5 to power a multi-agent framework capable of complex reasoning and autonomous decision-making.
• The Logic: We implemented a recursive Reflexion Loop where a Planner Agent drafts potential routes and a Critic Agent audits them.
• The Math: The Critic Agent evaluates recovery paths by calculating a weighted efficiency score $S$:

$$S = w_{time}(T_{delta}) + w_{cost}(C_{cost}) + w_{carbon}(CO_2)$$

The system optimizes this formula to find the detour that minimizes delays, extra costs, and environmental impact.

• The Visualization: The frontend was built using React and Vite, integrating Three.js to create an interactive 3D globe that visualizes cargo paths and disruption zones in real-time.
• The Infrastructure: We deployed the backend on Google Cloud Run and used Firebase Firestore to synchronize the AI's internal "thoughts" with the user dashboard instantaneously.

Challenges We Faced

• The "Hallucination" Barrier: Early in development, the AI would occasionally suggest geographically impossible routes. We solved this by implementing Gemini Tool Calling, forcing the agent to verify every waypoint through real-world Geocoding APIs before finalizing a plan.
• Data Latency: Official news feeds are often hours behind real-world events. This led us to build the Public Sentinel PWA, allowing crowdsourced "ground truth" reports from drivers and port workers to be ingested directly into the AI's reasoning engine.
• State Management: Coordinating the simultaneous logs of multiple AI agents while maintaining a high-performance 3D UI required a robust architecture to prevent lag and ensure a smooth "Command Center" experience.

What We Learned

Building CHRONOS taught us that the future of logistics isn't just about better data, but about autonomous agency. We learned how to move beyond basic prompt-engineering into complex agentic workflows where models critique their own logic to ensure operational safety. Most importantly, we discovered that sustainability can be treated as a hard mathematical constraint, making carbon-aware routing a standard feature rather than an afterthought.

Built With

• agentic-ai
• gsap
• javascript
• node.js
• openai-sdk
• three.js