Demo: Generate Accurate Product Recommendations

Video transcription

Disclaimer: Our transcriptions are generated automatically and may contain errors. They are provided for informational purposes only and should not be considered a substitute for watching the video itself. We cannot guarantee the accuracy, completeness, or usefulness of any transcription and accept no liability for any loss or damage resulting from its use. We recommend watching the video to ensure its accuracy.

I'm going to show you a couple of examples of how AI actions come into play. So this is a customer self help tree. Obviously it could also be an agent. It does not matter where it gets displayed, but I'm going to navigate to terms and conditions, click over here.  And in here, I'm going to give the end user the ability to ask a question and then feed that question into a large language model.

and then that large language model will be looking at this document over here. Which is basically the terms and conditions for this fictitious company insurance for all. and  and then spit out an answer.  I can ask anything. Do you cover cars? It's going into AI actions.

Looking against that document and spitting out an answer. The insurance policy does not cover cars. That is correct.  it pertains specifically to accidental damage and mechanical and electrical breakdown cover for insured items purchased from Amazons,  but does not cover cars. All right, we can ask another question.

I was in France  when I  broke  my phone.  Okay, and then ask that. Again, we're going, we're using actions to go tap into,  the large language model where we're looking through the document with, with context. And then the answer is that even if I'm in France,  it might still be covered by the insurance.

Even for instances where I'm traveling outside the country of residence, right? So that's one example.

Okay. So here's another example. I've, gone down the path of a tree and I get to, this place, after putting my, claim number. so now,  there's information about the user. The demographics or products that were bought because  they gave the claim number and then we tapped into a system and retrieve information about who's using Zingtree right now.

we were surfing actually resurfacing information. Right now from the database about the shipping address. Okay. So now click on continue and there's going to be a piece of AI actions into that answer. If you look specifically at the second piece here, this second sentence,  did you know that insurance for all offers policies for EV chargers?

And the reason why we're offering that as an output is because we made a business decision that in this case,  we think that if.  Somebody has two or more Apple products, then they're most likely an electric vehicle owner, right? And so let's let them know that we also cover EV chargers, right? Now, the business rules can be different, but in this case,  this particular user does have Two or more Apple products that are being insured by this insurance for all  company.

So that's, AI actions number two, and then AI action number three would be around,  a special offer just for you. Okay. So if I can click on next in this case, again, AI actions. Taps into a repository of available products, right?  that are, available for insurance. And, the rules here are just basically based on, demographics.

And we only have, age and gender, right? But based on age and gender. We're recommending a product that they think,  that they could get a discount on because maybe insurance for all has, also sells products or has partnerships with various product companies, and we're generating this output out of potentially thousands of different products that are being offered, but based on the large language model and how we define,  How we search against it, in this case, just age and gender, but it could be other things.