Intellisupply

View of digital twin maker
View of dashbaord with real time notifications about there supply chain relavent info
View of simulation section
View of form from where user can run different simulation of scenarios and get mitigation staergy
View of Simulation result section which comes after running the simulation
gives detailed finding about the simulation
When user clicks on mitigation strategy and see the strategy how to mitigate the probable problems ( preventive approach )

About IntelliSupply

In today’s volatile world, supply chain disruption is no longer an "if" but a "when," costing businesses trillions of dollars in lost revenue. Traditional management tools are reactive, leaving teams to clean up disasters rather than prevent them. We built IntelliSupply to change that paradigm, creating a proactive intelligence platform that empowers businesses to anticipate, simulate, and strategize against disruptions before they strike.

Our Inspiration: Connecting the Dots

Our journey began not with code, but with a headline: "How Water Scarcity Threatens Taiwan's Semiconductor Industry." The article revealed a slow-burning crisis with global implications. As we dug deeper, we realized a startling truth: the warning signs—from climate reports to local news and infrastructure alerts—were all there, scattered across the internet. The problem wasn't a lack of data, but a lack of a system intelligent enough to connect the dots. This was our "aha" moment. We were inspired to build an AI agent that could act as a digital watchtower, tirelessly scanning the horizon for these faint signals and translating them into clear, actionable warnings.

How We Built It: An Ecosystem of AI Agents

IntelliSupply is a next-generation platform built on a foundation of agentic AI. At its core, the system allows users to create a dynamic Digital Twin of their supply chain using an intuitive, node-based editor powered by Next.js and React Flow.

This entire experience is supercharged by a powerful AI Copilot, built using CopilotKit. This allows users to conversationally build, edit, and analyze their supply chain with simple text commands, turning complex manual tasks into effortless interactions. Also our general Copilot lets user navigate betweeen pages asks him about there digtial twins nodes , any relevant real time info etc .

The real magic happens in our backend, which is a multi-agent system designed for autonomous intelligence:

The Info Agent acts as our intelligence-gathering arm. It uses Tavily to perform deep, contextual searches on every node and edge in a user's supply chain, looking for real-time risks.
The Scenario Agent leverages Mem0 for long-term memory, recalling past risk patterns to generate highly relevant "what-if" disruption scenarios for users to simulate.
The Impact Assessment Agent quantifies the ripple effects of a simulated disruption, calculating financial impacts, operational delays, and generating cascading failure maps to visualize the full extent of a crisis.
The Strategy Agent uses Tavily for extensive research to build board-ready mitigation plans based on the simulation's impact assessment

All of these agents are powered by Google's Gemini for their core reasoning capabilities and use Supabase for data persistence and Redis for caching.

What We Learned

Building IntelliSupply was an incredible learning experience in agentic design. We learned that the true power of AI agents lies not in a single model, but in a well-orchestrated ecosystem. Integrating Mem0 taught us that providing an agent with memory transforms it from a simple tool into a learning partner that improves over time. Using Tavily showed us that giving an agent high-quality, real-time access to the web is like giving it eyes and ears, making its insights timely and grounded in reality. Finally, CopilotKit demonstrated how we can seamlessly embed these powerful backend agents directly into the user interface, making complex AI capabilities feel natural and intuitive.

Challenges We Faced

Our biggest challenge was time. We discovered the hackathon later in the cycle, which meant we had to be extremely focused and strategic with our development. This forced us to prioritize the core agentic loop over some UI polish we would have loved to add. Additionally, condensing the depth and breadth of IntelliSupply—from the digital twin editor to the simulation engine and strategy boards—into a concise 3-minute video was a significant challenge in itself.

Learn More About Our Architecture

We've documented our entire agentic architecture in detail. To dive deeper into how our system works, please check out our full technical blog post: Read About Our Architecture on CodeWarnab.in which has videos diagrams in detailed manner