Adapting Voice Assistance for Business: The Run
In our previous post, we discussed how to integrate Oracle JD Edwards with the Alexa voice interface to improve business efficiency with the help of voice input. In this blog post, I’ll explain the process and materials our team used to develop our first Alexa Skill with Amazon Web Services (AWS) Lambda.
Alexa Skill Development
After our initial Alexa Skill development, we saw a need to standardize the skills development process by following same methodology to reduce the development time. From there, we created a playbook to use as a one-stop guide for all the references needed for Alexa Skill development.
Since Alexa Skills and the Amazon Web Services (AWS) Lambda function development environment didn’t require additional hardware setup other than installing the Alexa Skills software development kit (SDK), the team directly jumped into the development with the playbook as a reference. Once we built the Accounts Payable Skill, we were able to demonstrate the following capabilities of Alexa:
- Custom slots to accept only predefined values (like month names)
- Ability to pass multi-word values from the voice interface (like vendor names)
- Saving the user’s selection (similar to session management in a web application)
- Ability to respond intelligently as per user selection (like responding differently By Vendor or By Company selection)
- For the same user utterance, executing a different intent depending on user selection
- Gracefully handling the situations where the intent is not defined in the Lambda function
- Built-in intents implementation (such as “Help,” “Yes,” “No”)
- Gracefully exiting user’s session at any point
For any system, usability places a vital role to receive by the user. During the development-testing iteration scope to improve the Alexa voice interface the team identified that some of the utterances defined were very lengthy and the Alexa voice interface could not process them as expected. The utterances defined in the Alexa Skill had to be fine-tuned to be shorter and more direct. Since there is no visual interface for the user, the Lambda function developed in such a way that it would repeat the user’s selection back to them as a confirmation before processing.
Spreading the Wings
With the expertise gained from this experiment, we were able to extend JD Edwards with Alexa voice assistance to different verticals such as Accounts Receivable, Sales Orders, and Inventory & Procurement. The Alexa voice interface could then offer assistance once prompted by simple instructions.
To conclude: armed with a clear separation of responsibilities, an in-house playbook, the agile development approach, and configurable templates, our JDE development team reproduced an Alexa Skill in a timely manner with adherence to standard practices and provided an intelligent service that can encourage business productivity.
There’s more to explore at Smartbridge.com!
Sign up to be notified when we publish articles, news, videos and more!