The latest articles on Forem by Mudigonda Venkata Gopi Jayaram (@jayaram9196). https://forem.com/jayaram9196 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3284734%2Fd98c6242-b0d4-4e3d-a959-11387f469567.png https://forem.com/jayaram9196 en Mudigonda Venkata Gopi Jayaram Sun, 22 Jun 2025 13:55:35 +0000 https://forem.com/jayaram9196/smart-supply-ai-32li https://forem.com/jayaram9196/smart-supply-ai-32li <h2> 🚀 SmartSupply AI: Multi-Agent Supply Chain Intelligence using Google ADK </h2> <blockquote> <p>Built for the <a href="https://googlecloudmultiagents.devpost.com/" rel="noopener noreferrer">ADK Hackathon</a> using Google’s Agent Development Kit (Python).</p> <p>🏷️ #adkhackathon #multiaagents #GoogleCloud</p> </blockquote> <h2> 🌟 Inspiration </h2> <p>The complexity of global supply chains — with their diverse geographies, regulations, and dependencies — inspired us to build <strong>SmartSupply AI</strong>, a <strong>multi-agent system</strong> designed to automate logistics intelligence and optimization. Our goal was to make supply chain decision-making more <strong>data-driven</strong>, <strong>resilient</strong>, and <strong>real-time</strong>.</p> <h2> 🤖 What It Does </h2> <p>SmartSupply AI is a <strong>multi-agent system powered by LLMs</strong>, purpose-built to streamline and simulate real-world supply chain logistics.</p> <p>It allows users to:</p> <ul> <li>🧭 Query global logistics routes (shipping, customs, trucking, warehousing).</li> <li>⚠️ Score risks based on delays, compliance, bottlenecks.</li> <li>🔁 Recommend supply chain optimizations for cost, time, and risk.</li> <li>📝 Collaboratively generate structured insights from agent orchestration.</li> </ul> <p>✅ Supports <strong>CLI</strong> and <strong>Web UI</strong><br><br> 🌍 Live Demo: <a href="https://demoagent-1.onrender.com" rel="noopener noreferrer">SmartSupply AI Web App</a></p> <h2> 🏗️ How We Built It </h2> <p>We used the <strong>Python version of Google’s Agent Development Kit (ADK)</strong> to orchestrate multiple agents:</p> <ol> <li> <strong>Fetch Agent</strong> → Pulls contextual data based on prompts.</li> <li> <strong>Risk Agent</strong> → Uses domain heuristics to assess route/product risks.</li> <li> <strong>Optimization Agent</strong> → Suggests improvements to reduce delays and costs.</li> <li> <strong>Report Agent</strong> → Summarizes results in readable formats.</li> </ol> <p>We deployed the system on <strong>Render</strong> and provided a <strong>CLI and web interface</strong>.</p> <h2> ⚙️ Architecture </h2> <p><a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fhpeanlcbqyybuaadpw3r.png" class="article-body-image-wrapper"><img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Farticles%2Fhpeanlcbqyybuaadpw3r.png" alt="smartsupplyai" width="800" height="106"></a></p> <h2> 🧠 Sample Prompts for Real-World Testing </h2> <p>Try these to experience the full power of agent coordination:</p> <ul> <li>📦 <em>Smartphones from Shenzhen to Chennai</em> </li> <li>🖥 <em>Monitors from Taiwan to Noida</em> </li> <li>💊 <em>Pharma APIs (Cold Chain) from Germany to Bhiwandi</em> </li> <li>👗 <em>Textiles from Dhaka to Kolkata (via Petrapole)</em> </li> <li>🥤 <em>Packaged Beverages from Pune to Ahmedabad</em> </li> </ul> <p>Copy/paste these into the app to test real-world multi-agent analysis.</p> <h2> 🧗 Challenges We Faced </h2> <ul> <li>Understanding agent chaining and memory persistence with ADK.</li> <li>Managing external credentials (Google API keys, service accounts).</li> <li>Handling cold-chain pharma routing complexity.</li> <li>Deploying the system with minimal latency on Render.</li> </ul> <h2> 🏆 Accomplishments </h2> <ul> <li>Successfully deployed a real, working <strong>multi-agent logistics intelligence platform</strong>.</li> <li>Created reusable components using <strong>ADK’s SequentialAgent</strong> pipeline.</li> <li>Integrated LLM tools and domain-specific logic without fine-tuning.</li> </ul> <h2> 📘 What We Learned </h2> <ul> <li>Deep dive into the ADK’s modular agent orchestration capabilities.</li> <li>Efficient multi-agent task delegation and result consolidation.</li> <li>Leveraging language models for <strong>structured B2B tasks</strong>, not just chat.</li> </ul> <h2> 🔮 What's Next </h2> <ul> <li>Add PDF export (via GCS Artifacts).</li> <li>Plug in <strong>real logistics APIs</strong> (e.g., Freightos, Portcast).</li> <li>Enable <strong>visual dashboards</strong> (charts, maps).</li> <li>Extend to include <strong>carbon emissions modeling</strong>.</li> </ul> <h2> 🛠️ Built With </h2> <ul> <li><a href="https://www.python.org/" rel="noopener noreferrer">Python</a></li> <li><a href="https://github.com/google/agent-development-kit" rel="noopener noreferrer">Google Agent Development Kit (ADK)</a></li> <li> <a href="https://render.com/" rel="noopener noreferrer">Render</a> (Hosting)</li> <li><a href="https://dev.toPlanned%20for%20future">Google Cloud Storage (GCS)</a></li> <li><a href="https://github.com/jayaram9196/DemoAgent" rel="noopener noreferrer">GitHub</a></li> </ul> <h2> 📦 GitHub Repository </h2> <p>🔗 <a href="https://github.com/jayaram9196/DemoAgent" rel="noopener noreferrer">jayaram9196/DemoAgent</a></p> <p>Clone and test using:</p> <div class="highlight js-code-highlight"> <pre class="highlight plaintext"><code> bash git clone https://github.com/jayaram9196/DemoAgent.git cd DemoAgent python3 -m venv venv source venv/bin/activate pip install -r requirements.txt adk web # to start the web interface </code></pre> </div> ai python googleadk adkhackathon