Inspiration
In chat commerce today, to get customer delivery location is same way as they put the address in text box when people do checkout on the website. Meanwhile it is hard to determine point location if it only using address text from customer. Many logistic companies in here have to convert it to geo location and most of the result was not precise.
In other hand, we have similar district name area and we cannot rely on postal code only.
Getting exactly point location (known as LongLat) from customer has big benefit to helps company create billing shipping cost accurately, do delivery faster and precisely, assigned courier job automatically and more.
What it does
It is a Messenger bot that have knowing conversation about place and allows you to pin point location. You will be asking about your specific place that you want to delivery. It will validate your place and if the place was not found other suggestion will be sent to.
How we built it
We used Amazon Lex as bot engine conversation. There are one intent (gettingLocation) and two slots (deliveryPlace, pinLocation). To elicit deliveryPlace we created lambda function which will determine whether slot need a confirmation with other suggestion, it has been accepted by customer, or it was rejected.
After that pinLocation will be elicited using map interface. We shows map area based on deliveryPlace slot value so that it make easy for customer pin their location then. We need No-SQL database to store values to keep context conversation between webview Messenger and Lex bot.
We built map interface using ionic angular and show it through webview messenger and it will follow UI component on different device.
Challenges we ran into
It is not easy to make interactive conversation from our idea. We have bad luck to use Messenger Extension SDK on getContext() function. Nothing success callback return from it. Instead of using signed_request value we use owned token from Dynamodb.
Accomplishments that we're proud of
We don't have deep experience using node js and angular but we tried our best to make this serverless design architecture.
What we learned
Facebook Messenger has a powerful platform chat and it easy to integration with.
What's next for Agent Owl
It will help logistic company and inspiring to another one.
Built With
- amazon-cloudfront-cdn
- amazon-dynamodb
- amazon-lex
- angular.js
- api-gateway
- facebook-messenger
- google-geocoding
- ionic
- lambda-function
- node.js
- typescript

Log in or sign up for Devpost to join the conversation.