Amazon Lex FAQ (Frequently Asked Questions)

Here is our Amazon Lex FAQ.

First Up In Our Amazon Lex FAQ – What Is Amazon Lex?

Amazon Lex is part of omnichannel cloud contact centre that helps companies provide superior customer service across voice and chat at a lower cost than traditional contact centre systems.

 

Amazon Lex Logo

 

For more information on how Route One Connect can assist you in your project, click here.

 

What is Amazon Lex FAQ

 

Secondly, Does Amazon Lex get more intelligent over time?

Absolutely. Because of artificial intelligence and Amazon Lex’s machine learning capabilities the performance will get better over time. However, to achieve the performance you would expect in a Contact Centre you must have a deep understanding of contact centre’s, Speech and behaviour understanding. Relying on AI and ML to meet your performance and CX goals will be a costly mistake.

 

What is Amazon Lex FAQ Separator

 

Do AWS Carry Out Speech Science And Data Science Work On Your Lex bot Or Is That Up To The Customer?

AWS only support the Amazon Lex product. What goes in and what is configured in Amazon Lex is down to the customer or AWS partner.

 

What is Amazon Lex separator

 

Do You Need To Be A Speech Scientist To Build And Design An Amazon Lex Bot?

This is a great question. Currently there is a flood of AWS engineers now able to create Amazon Connect and Amazon Lex services because they have been working with AWS for 10+ years. However, without a deep understanding of Contact Centres, how Natural Language Understanding (NLU) works and even cognitive behaviour theory then sadly the solution will not work well. And this is important! Above all, if you get it wrong, you will destroy the customer experience. Likewise, and the trust of your customer!

 

separator

 

Where Can I Learn More About How Amazon Connect Works?

Here is a link to AWS Lex website where you can get more information on Amazon Lex. Click here to learn more.

 

separator

 

What Is An Utterance?

Utterances are the words spoken or written that is to be understood by Amazon Lex. For Instance: if an Amazon Lex Conversational AI solution asks

“Welcome to Route One Connect, how can I help you today?”

Customer: “I’d like to update my electricity meter reading”

The utterance would be “I’d like to update my electricity meter reading”

 

separator

 

Does the sample utterance list need to be explicit?

No, this is where artificial intelligence comes in. If in your sample utterance configuration you specified the below configuration, the AI algorithm in Amazon Lex would mean that if i said something similar that the Intent of ‘PaymentTransfer’ would still be mapped.

{Intent = PaymentTransfer, Sample Utterance = “I’d like to transfer [slot: amount] from my savings account to my ISA account”}

This is important when we consider the impact of not recognising an utterance correctly. For instance, if the utterance is mis-recognised then design might say transfer the call\chat to a general skill. This would then impact first call resolution, average handling time and increase internal transfers. These three are synonymous with a poor customer experience so it’s vital there is a focus on correct intent matching.

This is a great example of how technology has moved on from legacy platforms. With legacy platforms the list of utterances is an explicit list and if the customer says anything different to what has been configured then a no match is thrown.

 

separator

 

How has Amazon Lex moved the technology space on from legacy platforms?

  • Amazon Lex has a fulfilment option. This essentially allows the developer to call external APIs, manipulate the data received and returned. Incredibly powerful when we think about reporting and CRM integration. Legacy platforms just do speech recognition only, API calls would have to exist completely external from the speech platform.
  • Sample utterances do not need to be explicit with Amazon Lex to be recognised. The AI and ML functionality is what allows the Lex bot to perform better on a limited list on sample utterances. On the other hand, the richer the data you feed Amazon Lex the better it will perform.

 

 

In Conclusion, we hope you enjoyed reading our Amazon Lex FAQ, if you did these other blogs may be of interest:

Amazon Lex Blog

Beginners Guide To Conversational AI

Amazon Connect Best Practice – Part 1

Amazon Connect FAQ (Frequently Asked Questions)

 

Finally, for more information on how Route One Connect can assist you in your project, click here.