Inspiration
Cafés and retail outlets often struggle with fragmented digital ordering systems, long queues, and limited personalization. We were inspired to create a solution that feels like talking to a friendly barista—simple, fast, and available across WhatsApp, mobile, and web.
What it does
Our AI-powered multichannel ordering system lets customers order via WhatsApp, mobile app, or website using natural text or voice. A smart AI assistant powered by Amazon Bedrock understands the order, checks availability, suggests alternatives, manages queues, and triggers payment links via Revolut. On the business side, shop owners upload menus instantly, track live orders, and get real-time analytics powered by Amazon SageMaker.
How we built it
AI Agent: Amazon Bedrock models for natural language understanding
Backend: AWS Lambda, API Gateway, DynamoDB, S3, CloudFront
Analytics: Amazon SageMaker for trends and demand forecasting
Messaging: WhatsApp Cloud API / Twilio integration
Payments: Revolut API with optional Stripe fallback
Frontend: React (web), Flutter (mobile), and a self-serve menu management portal
Challenges we ran into
Handling real-time availability across multiple channels
Training AI prompts to handle corrections, substitutions, and upsells naturally
Ensuring low-friction onboarding for shops with no technical background
Integrating secure payments and confirming them instantly in the order flow
Accomplishments that we’re proud of
Built a no-code menu upload system—shops paste a CSV or text, and the AI auto-configures itself
Achieved multilingual support for both voice and text ordering
Seamlessly connected WhatsApp → AI → Payments → Shop Dashboard in a single flow
Delivered real-time analytics with predictive insights on peak times and repeat customers
What we learned
Conversational AI requires context awareness—we had to refine prompts to handle ambiguous or casual customer phrasing
Shops want simplicity over customization—automating setup is as important as AI accuracy
Data analytics can be a game-changer for small shops when made accessible and visual
Built With
- amazon-dynamodb
- amazon-web-services
- api
- appsync
- bedrock
- social-media-messaging