The latest articles on DEV Community by Shobana Sreedharan (@shobana_sreedharan_f776bb). https://dev.to/shobana_sreedharan_f776bb https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4006959%2F3618059a-fe26-4e2c-9061-6e34b2435b43.png https://dev.to/shobana_sreedharan_f776bb en Shobana Sreedharan Sun, 28 Jun 2026 20:20:22 +0000 https://dev.to/shobana_sreedharan_f776bb/how-i-built-shopsmart-ai-with-aws-dynamodb-and-vercel-for-the-h01-hackathon-tags-aws-vercel-298 https://dev.to/shobana_sreedharan_f776bb/how-i-built-shopsmart-ai-with-aws-dynamodb-and-vercel-for-the-h01-hackathon-tags-aws-vercel-298 <p>I created this project for the H01 AWS + Vercel Hackathon (#H0Hackathon).</p> <h2> What I Built </h2> <p>ShopSmart AI is a conversational shopping assistant that helps users <br> plan events and DIY projects by recommending the right products instantly.</p> <p>Try it live: <a href="https://shopsmart-ai-build.vercel.app" rel="noopener noreferrer">https://shopsmart-ai-build.vercel.app</a></p> <h2> The Problem </h2> <p>Planning an event or tackling a DIY project means spending hours <br> searching across multiple sites. ShopSmart AI eliminates that friction <br> — describe what you need in plain English and get a smart product catalog instantly.</p> <h2> Tech Stack </h2> <ul> <li> <strong>AWS DynamoDB</strong> — product catalog (40 items, 8 categories, composite key: category + productId)</li> <li> <strong>Vercel</strong> — frontend deployment and AI Gateway</li> <li> <strong>Next.js + TypeScript</strong> — frontend framework</li> <li> <strong>Vercel AI SDK</strong> — streaming chat with 5 custom tools</li> <li> <strong>GPT-4o-mini</strong> — AI recommendations</li> </ul> <h2> How It Works </h2> <ol> <li>User describes their event or DIY project in the chat</li> <li>AI asks clarifying questions</li> <li>User confirms with "ok"</li> <li>AI fetches real products from DynamoDB</li> <li>Catalog populates with categorized product recommendations</li> </ol> <h2> DynamoDB Architecture </h2> <p>The table <code>shopsmart-db</code> uses a composite key:</p> <ul> <li>Partition key: <code>category</code> (e.g. Decorations, Electrical, Plumbing)</li> <li>Sort key: <code>productId</code> </li> </ul> <p>On first query, AI recommendations are saved to DynamoDB. <br> Repeat queries serve from the database — faster and cheaper at scale.</p> <h2> What's Next </h2> <ul> <li>Affiliate integration with Amazon and Target</li> <li>User accounts with persistent shopping lists</li> <li>Mobile app</li> <li>B2B white-label API for retailers</li> </ul> <h2> Conclusion </h2> <p>Building with AWS DynamoDB and Vercel was a great experience — <br> the Vercel native AWS integration via OIDC tokens made credential <br> management seamless in production.</p> aws vercel nextjs hackathon