Tuali - Ruta Directa
Inspiration
Every week, hundreds of B2B clients receive products they didn't order and nobody tells them. We read Arca Continental's challenge and recognized a real operational pain: the substitution happens in seconds at the CEDIS, but the impact hits the client hours later at their door. We built Ruta Directa to close that gap.
What it does
Ruta Directa is a delivery management platform with three specialized interfaces: client, admin, and delivery driver, sharing one real-time backend. Unlike solutions that only react once something goes wrong, we tackled the problem at every stage of the delivery process, from the moment an order is placed all the way through completion.
Before dispatch, Gemini AI predicts which products are at risk of running out and tells the admin exactly what to restock, so the substitution never happens in the first place. During the order, clients get proactive in-app notifications when the AI detects they're due for a reorder based on their history, and can repeat a previous order in one click. If a substitution does occur, clients are notified via WhatsApp immediately. And at delivery, drivers report incidents on the spot so nothing falls through the cracks.
The admin can reorder inventory directly from the prediction dashboard, closing the loop before the truck ever leaves.
How we built it
Frontend in React 19 + TypeScript with Tailwind, Framer Motion, Recharts for analytics, and Leaflet for interactive route maps. Backend in Node.js + Express with MongoDB on Railway. AI predictions powered by Google Gemini. WhatsApp notifications via Twilio. We built separate role-based dashboards that share one synchronized backend, with JWT auth and automatic role detection on login.
Challenges we ran into
Getting the three roles to reflect state changes in real time without polling was harder than expected. Training Gemini to produce useful inventory predictions from historical order data, not just generic responses, required significant prompt engineering.
Accomplishments that we're proud of
The substitution prediction feature works end-to-end: Gemini analyzes historical data, flags products by urgency, proposes a specific replacement product, and the admin can trigger a restock in three clicks. The WhatsApp notification flow fires automatically when a driver marks an incident, no manual step required.
We also care deeply about accessibility. Not every B2B client has the time or ability to navigate a web app mid-workday. That's why we integrated ElevenLabs so clients can place orders entirely by voice, no typing needed. And for delivery updates, we built a conversational chatbot using Gemini that clients can query at any time to get the current status of their order in plain language.
What we learned
Logistics problems look simple on the surface and get complicated fast once you model the real actors involved. We learned that the most valuable thing you can build for an operations team isn't a beautiful dashboard, it's removing one decision they have to make manually under time pressure. We also learned that accessibility isn't a nice-to-have in B2B logistics: a client running a restaurant or a corner store doesn't stop their day to check an app.
What's next for Ruta Directa
Digital signature capture at delivery. Expanding the voice ordering experience to WhatsApp so clients can place orders conversationally without any app at all.
Built With
- cloudinary
- elevenlabs
- express.js
- framer-motion
- git
- github
- google-gemini-ai
- leaflet.js
- mongodb
- node.js
- railway
- react-19
- recharts
- tailwind-css
- twilio-whatsapp-api
- typescript
- vite
Log in or sign up for Devpost to join the conversation.