Inspiration
At VGW, we pride ourselves at providing world-class customer service to our customers.
As the only online casino in the US where players can legally cash out their winnings through sweepstakes technology, VGW's installed customer base is growing steadily and we love it when our customers win!
Currently, when VGW's customer's request a cash out of their winnings, they need to contact our award-winning customer service team to check on the status. Although VGW's customer service team strives to respond to these queries as quickly as possible, there is a delay as the inquiry goes through the service team to be processed.
VGW hopes to be able to use lex to be able to help customers self-service cash out status checks to improve the responsiveness of VGW's core business offering to the customer. Customer should be able to query the status of their wins as soon as, and as flexibly as they require.
What it does
The VGW Cash out bot allows customers to self-service their cash out request status. It currently does so in two ways:
- visiting the site https://lex.nerdcruft.net and interacting with the bot conversationally via spoken word
- via SMS
In order to support conversational expressions from the customer, the cash out bot is able to fuzzy search through cash out dates so that we might be able to locate a cash out. For example "last week" or "today" instead of a hard date "YYYY-MM-DD"
How we built it
The cash out bot's Lambda handler is written in python, and queries an endpoint with customer cash out statuses.
When a customer uses the webpage, they interact with a Amazon Lex, which prompts them conversationally to provide enough information in order to query their cash out status.
With the provided information, the Lambda is able to search through the provided list of cash outs for the best match and return a result.
Challenges we ran into
Service orchestration for a small team new to aws felt artificially complicated and we spent a fair fraction of time on setup rather than product creation. it was nontrivial to setup a demonstration system for a relative newcomer. debugging and automated testing is more difficult to achieve on a system based around SaaS as compared to local deployment.
Lex doesn't support currency as a data type! Come on Amazon!
Accomplishments that we're proud of
This project allows VGW to experiment with reaching the less-technologically savvy segments of VGW's customer base.
It also expands VGW's customer touchpoints to include SMS as a channel.
The bulk of this code was hacked together by a Business Analyst and not a programmer. Whilst the learn curve for getting AWS services set up was steep initially, once we got there, it became a lot easier.
What we learned
- Lambda is cool.
- IAM is a pain(*).
- Lex is neat, and is going to be AWESOME.
- Don't expect that example projects on github to work - sometimes they are outdated.
- People really really really like talking to their computers.
(*) opinions expressed are my own and do not represent that of my employer
What's next for VGW Cashout Bot
- Exploring other customer self service intents such as Self-Exclusion Requests, Play enquiries etc.
- Exploring porting to mobile (Media Devices doesn't work on mobile safari or android)
- Extension and integration with Zendesk to enable Customer Service team follow up
Built With
- javascript
- lex
- lightsail
- python
- soundex
Log in or sign up for Devpost to join the conversation.