The latest articles on DEV Community by Rozeena (@rossycoder). https://dev.to/rossycoder 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%2F4181599%2F51f335d4-28da-4fa6-92f8-846f6316f1df.png https://dev.to/rossycoder en Rozeena Tue, 13 Oct 2026 08:20:17 +0000 https://dev.to/rossycoder/i-built-an-ai-brain-for-car-dealerships-that-never-forgets-a-customer-36ch https://dev.to/rossycoder/i-built-an-ai-brain-for-car-dealerships-that-never-forgets-a-customer-36ch <p>Every car dealership has the same problem: a customer calls, gets promised a callback, and nobody calls back. The customer buys from a competitor. Dealership staff aren't lazy — they're drowning in sticky notes, spreadsheets, and forgotten promises.</p> <p>So I built <strong>Dealership Brain</strong> — a graph-powered AI assistant that remembers every customer, every vehicle they looked at, every promise made to them, and every service visit. Nothing falls through the cracks.</p> <p>🔗 <strong>Live demo:</strong> <a href="https://dealership-brain.vercel.app" rel="noopener noreferrer">https://dealership-brain.vercel.app</a><br> 💻 <strong>Code:</strong> <a href="https://github.com/rossycoder/dealership-brain" rel="noopener noreferrer">https://github.com/rossycoder/dealership-brain</a></p> <h2> What it does </h2> <ul> <li> <strong>AI chat (English + Roman Urdu)</strong> that answers from the dealership's real inventory and the customer's history</li> <li> <strong>Callback SLA tracker</strong> — every promised callback gets a deadline; miss it and it escalates to the manager automatically</li> <li> <strong>Promise tracker</strong> — "we'll courier your spare key" becomes a tracked commitment with a due time, not a forgotten sentence</li> <li> <strong>Visual customer graph</strong> — see a customer's vehicles, tickets, conversations and promises as one interactive map</li> <li> <strong>Smart Match</strong> — when a new car arrives, it finds the customers whose budget and history fit, and drafts a WhatsApp follow-up for each</li> <li> <strong>Trade-in valuation, test-drive booking, EMI calculator, service reminders</strong> — the everyday dealership toolkit in one place</li> </ul> <h2> How it works </h2> <p>The heart is a <strong>knowledge graph</strong> (FalkorDB, with an in-memory fallback for serverless): customers, vehicles, tickets, callbacks and promises are nodes; everything the business "knows" is an edge. AI agents read from and write to that graph, so the system literally builds a memory of the business. A trace panel shows every decision, query and graph write — no black box.</p> <p>Built with FastAPI + vanilla JS on Vercel (free tier). I built this for the FalkorDB Graph Hacks hackathon.</p> <h2> Try it </h2> <p>Open the demo, pick the Manager view, and simulate a missed callback — watch it escalate. Then ask the chat for "an SUV under 50 lakh" in Roman Urdu.</p> <p>I'd love feedback from anyone who has worked in car sales: what would your dealership need this to do?</p> ai hackathon webdev showdev