Skip to content

Latest commit

 

History

History
106 lines (65 loc) · 6 KB

File metadata and controls

106 lines (65 loc) · 6 KB

LCA AGENTASSIST Demo


Part 1: AgentAssist gathers some information for later use - eg credit card type and caller’s first name.


Agent: Hello! Thanks for calling ACME Bank. My name is Oliver. How can i help you?
Caller: Hi. I’m calling about my Rewards card.
Agent: Excellent.. give me a second to look up your customer account..

AGENTASSIST: Ask caller to state their first name only, for verification purposes. It should be ‘Bob’ per callerId lookup.

Agent: For verification purposes, can you please state your first name?
Caller: My first name is Bob.
Agent: one moment while we verify* that your name matches our records.*

AGENTASSIST: Suggested response: “Perfect! Hi Bob. How can i help you with your rewards-card?”

Agent: It matches - perfect! Hi Bob. How can i help you with your rewards-card?*


Part 2: AgentAssist retrieves information using the information gathered earlier (eg card type).. in this case by using conditional answer in QnAbot, but other mechanisms - such as Kendra redirect query filters or custom logic with Lambda hooks - could also be used.


Caller: I was wondering, what is the cash back rate for this card?
Agent: I’m looking that up for you now. It won’t take long.

AGENTASSIST: Suggested response: Yes, I see here that the rewards card has a 2% cashback rate on your everyday spending. Did that answer your question? Anything else I can help with today?

Agent: Yes, I see here that the rewards card has a 2% cashback rate on your everyday spending. Did that answer your question? Anything else I can help with today?


Part 3: AgentAssist can guide agent through an online transaction, prompting agent to collect the needed information. Potentially AgentAssist could fully automate transaction fulfillment.


Caller: Thanks, yes. Can i pay my card bill now while I’m on the phone?
Agent: Yes, absolutely, we can process your payment over the phone. Give me a second to set that up.

AGENTASSIST: Ask caller for last 4 digits of card number

Agent: Can you please confirm the last 4 digits of your card number?
Caller: Yes. It’s 1234
Agent: Got it - 1234.

AGENTASSIST: Confirm card balance (eg $75.34). Ask caller how much they want to pay ($50 minimum)

Agent: Great. I see that your card balance is $75.34. How much do you want to pay? There’s a $50 minimum payment.
Caller: $50
Agent: OK, I’m entering payment for $50.

AGENTASSIST: Ask caller if they want to pay from their checking or savings account

Agent: Which account do you want to pay that from.. Your Savings account, or your Checking account?
Caller: My checking account.
Agent: Your checking account - perfect. One moment please.

AGENTASSIST: All information captured.. Last 4 = 1234, Amount = $50, Account = Checking

Agent: Great, so I’ve processed payment of $50 from your Checking account. Is there anything else I can help you with?


Part 4: Caller asks a question where we have an Kendra FAQ answer in QnABot


Caller: Can you tell me what kinds of life insurance policies you have?
Agent: Yes, of course. I’m just pulling up our latest life insurance policy information now.

AGENTASSIST: Open https://en.wikipedia.org/wiki/Life_insurance: We offer many types of Term life insurance, and Permanent life insurance, including Whole life, Universal life, Endowment policies, Accidental death policies, and more.

Agent: We offer many types of Term life insurance, and Permanent life insurance, including Whole life, Universal life, Endowment policies, Accidental death policies, and more.


Part 5: Caller asks a question for which we don’t have an FAQ.. Demonstrates fallback to a Kendra query on unstructured docs from the index (in the case the crawled website)


Caller: Interesting. What is an endowment policy?
Agent: I'll tell you in one second!

AGENTASSIST: Amazon Kendra suggestions. If the cash value grows over time, the death benefits do too. If the cash value declines, the death benefit also declines. Option B policies normally feature higher premiums than option A policies. Endowments[edit] Main article: Endowment policy The endowment policy is a life insurance contract designed to pay a lump sum after a specific term (on its 'maturity') or on death. Typical maturities are ten, fifteen or twenty years up to a certain age limit. Source Link: https://en.wikipedia.org/wiki/Life_insurance

Agent: It's a life insurance contract designed to pay a lump sum after a specific term (on its 'maturity') or on death. Typical maturities are ten, fifteen or twenty years up to a certain age limit. Let me send you some information and let’s schedule a chat with one of our insurance specialists. Does that sound OK?


Part 6: Wrapup


Caller: That sounds great. Thanks.
Agent: Is there anything else I can help you with today?
Caller: Nothing else just now. You’ve been a big help. Thank you so much.
Agent: No problem. I am very glad to hear you say that I was able to help you today.

AGENTASSIST: Thanks Bob! We appreciate your business. Check out our website at ACMEBank.com to learn about all our other products, and feel free to call back if you have any additional questions. Have a great day.

Agent: Thanks Bob! We appreciate your business. Check out our website at ACMEBank.com to learn about all our other products, and feel free to call back if you have any additional questions. Have a great day.
Caller: Thanks. Bye.
Agent: Goodbye.