Menu
February 25, 2019 | By

AI IRL Podcast Episode 8: What to Consider When Designing a Virtual Assistant


Subscribe via iTunes, Spotify and more.

Creating chatbots is a frontier.

Fortunately, my guest Qing Cao, in addition to being a certified Lean Six Sigma black belt, is the leader of the customer-facing, emerging technology product development team at UPS.

On a recent episode of AI:IRL she joined me to unpack the different ways you can build a bot to fit your situation and defining your purpose and asking guiding questions for decision-making.

Building a Bot to Fit Your Situation

Chatbots or virtual assistants fall into two categories and your use-case should play a role in deciding between them.

Using a prebuilt service

There are many companies that offer services such that you don’t even need developers to build a bot.

You simply go into their system, write down decision trees, what you think the customer will ask, and what the responses should be.

This is a great option for less complex use cases.

Traditional way

In the traditional method of building a virtual assistant, you will need developers. But you can build it on Microsoft Bot Framework and leverage the Google assistant or Alexa platforms.

In this case, the developers are doing the heavy lifting of inputting the utterances and the answers.

This is the best option for more complex use cases or when your company wants to own IP.

Qing’s company has two different products leveraging both methods. She works primarily in the traditional developer route and there’s another team that uses a service to build another virtual assistant.

”It’s about figuring out the best way to build the chatbot to serve the customers.” – Qing Cao

Defining Your Purpose as a Guide for Decisions

So how do you go about making the business decisions involved in chatbot design?

Start by defining your purpose.

Here are some rudimentary questions whose answers should serve as a guide in the decision making:

  • Who’s your target audience?
  • What platforms are they on?
  • Are they likely to visit your website or download a mobile app?
  • Or should your bot live on a native platform like Google or Alexa.

”If you are considering building a chatbot, you have to start by defining your purpose.” – Qing Cao

Expanding to Other Platforms, Languages, and Regions

Expansion is key to any business’s growth.

Inevitably, successful businesses with virtual assistants will need to think about expanding to other platforms, languages, and regions.

Platforms

When building a bot, potentially you would build it multiple times for different platforms depending on the different requirements or limitations of each

Across Apple, WeChat, and Facebook Messenger, for example, there are commonalities in API, but there are also some dramatic differences.

You will need whole teams just working on keeping up with your integrations.

Languages

Nowadays translation technology is very mature, and it is not difficult to leverage for customer input.

However,  you have to control the output. This involves managing all the content you’re putting out in however many languages you want to support.

What some companies are doing is having the output for simple text interactions translated by AI while having larger more important articles and documents translated by human translators.

Regions

The last challenge is geographical concerns about data rules and privacy.

As you move to Europe, as you move to mainland China, the way you can access CPIs and data varies significantly.

So this is another thing to be thoughtful of as you design because one architecture in one place may not be able to be leveraged in other geographies.

Having a Test-and-See Mentality.

Qing’s last piece of advice is about the mentality to have in chatbot design. She emphasizes the importance of having fun and not being afraid to fail.

Building a chatbot is fun. And It is in trend right now. Have fun during the process. Test and learn. There are many possible solutions

Don’t be afraid to fail. There many bots that are failing out there right now. You need to learn from failure to make a better next version.” – Qing Cao

This AI discussion with Qing Cao was taken from our podcast. If you want more AI: In Real Life, check out our podcast on iTunes.

View transcript »

Ryan Lester
00:05 – 00:09
Hello, welcome, this is Ryan Lester host of AI in real life.
Ryan Lester
00:09 – 00:10
Welcome to this week’s podcast.
Ryan Lester
00:10 – 00:33
I have a really exciting guest this week a chain from UPS customer and customer technology marketing a chain currently leads the customer facing emerging technology product development team at UPS working with teams across multiple states of the US and in fact even several countries in Asia focusing on deliver better customer experiences prior to joining UPS.
Ryan Lester
00:33 – 00:38
She worked in global procurement and sourcing specifically with a number of Chinese vendors.
Ryan Lester
00:38 – 00:41
She’s a native Mandarin Chinese speaker.
Ryan Lester
00:41 – 00:47
We’ve up I won’t get into that today, but she also a certified with lean Six Sigma Black Belt.
Ryan Lester
00:47 – 00:51
So she brings that certainly operational view to the world of customer experience.
Ryan Lester
00:51 – 00:55
So Ching welcome really great to have you join the podcast this week.
Cao Qing
00:56 – 00:56
Thank you.
Cao Qing
00:56 – 00:57
Ryan.
Cao Qing
00:57 – 00:58
Very excited for today’s talk.
Cao Qing
00:59 – 01:00
Yeah, same here.
Cao Qing
01:00 – 01:09
I know you and I have spoken in the past on some panels and I know you bring a real wealth of knowledge in the world of chat Bots and Digital customer engagement.
Ryan Lester
01:09 – 01:19
So I’m excited to kind of dive in to the Waters of your experiences and what you’ve learned in your work in the world of chat Bots and virtual assistants sounds great.
Cao Qing
01:19 – 01:40
There’s a lot to talk about there certainly is so the first thing is, you know, people think about a chatbot and they obviously I think often think of the customer facing experience so the bot itself and that little window that the bot interacts with but in reality there’s quite a bit behind the scenes, you know, if you were to kind of open the curtain there’s a lot that’s going on in the background.
Ryan Lester
01:40 – 01:51
So what I want to spend some time with is in your experience and the work you’ve done with chatbots, let’s talk through some of the challenges of implementing a chatbot that you know, there’s a whole operation behind it.
Ryan Lester
01:51 – 01:55
I like to think about things is like a factory save a big Factory back there and there’s lots of
Ryan Lester
01:56 – 02:03
Says things like knowledge bases Ai, and I’ll pee the platform different languages and geographies.
Ryan Lester
02:03 – 02:07
So what are we start down the road of just talking about what are some of the core pieces?
Ryan Lester
02:07 – 02:10
You need to really Implement a chatbot?
Cao Qing
02:11 – 02:18
That’s a great question to start with for people who are considering building a chatbot.
Cao Qing
02:18 – 02:22
I would say to understand what are the ways you can go to chat Bots.
Cao Qing
02:22 – 03:10
Like what I understood as one is you have a lot of companies or vendors offer services that you don’t really need developers to build a box you basically go into their system right down decision trees and how how you think the quest of the customers will ask a certain question and then how using the response should be and then you build it that way or you can build it in the traditional way or the developers required way which you build it either on Microsoft bought Frameworks or you leverage Google or Lexus platforms, and then really developers are doing the heavy lifting of inputting the utterances and the answers and all that.
Cao Qing
03:11 – 03:12
That’s the first step.
Cao Qing
03:12 – 03:23
Yeah, and it sounds like that’s the route you guys went in the sense of you felt like for you the core competency was kind of building a lot of the core platform yourself off of some of these two kind of Technology platforms.
Cao Qing
03:23 – 03:32
Yeah, so we actually as in my company who actually has two different products leveraging both methods.
Cao Qing
03:33 – 03:49
So where I’m at whereon they Agile development team so the developer routes and there’s another team here that uses a platform or service building and the other virtual assistant.
Cao Qing
03:49 – 03:50
Well, that’s interesting.
Cao Qing
03:50 – 03:57
It’s actually it’s something we’ve seen as well where companies may take kind of two approaches where you know, maybe some of their use cases are more complex.
Ryan Lester
03:57 – 04:09
They want to kind of or own more the IP and then other cases it may be less complex and they choose to partner with someone who brings more of a complete solution or a more finished solution to bear.
Ryan Lester
04:09 – 04:10
So that’s something that we’ve seen elsewhere.
Ryan Lester
04:11 – 04:11
Yeah, absolutely.
Cao Qing
04:12 – 04:13
There’s pros and cons.
Cao Qing
04:13 – 04:15
I was safe in both cases.
Cao Qing
04:15 – 04:18
But this is the world we’re living in right now.
Cao Qing
04:18 – 04:22
Everybody’s trying to figure out the best way to serve the customers.
Cao Qing
04:22 – 04:23
So to chat box.
Cao Qing
04:23 – 04:34
So it’s this is I would say something to look into and figure out what fits your situation the best first and then understand the knowledge bases.
Cao Qing
04:34 – 04:42
Like if you’re doing two sets of things, how do you merge knowledge bases potentially or you don’t merge knowledge bases potentially.
Cao Qing
04:42 – 04:47
So yeah, that’s great and are there some business decisions or things that you looked at?
Ryan Lester
04:47 – 04:52
Was it the type of information housed or was it based on like business group?
Ryan Lester
04:52 – 04:57
How did you make some of those decisions when you’re saying where information should live and who should own those knowledge bases?
Cao Qing
04:58 – 05:04
Those I mean, that’s what that’s a really hard question.
Cao Qing
05:07 – 05:13
So it’s actually so it depends on where your ads in development.
Cao Qing
05:13 – 05:16
Some of the decisions were made early on some of the decisions.
Cao Qing
05:16 – 05:22
Some of the problems were fined after you already started separate ways.
Cao Qing
05:22 – 05:30
So it is hard to say but I would say if other companies are considering building a chatbot.
Cao Qing
05:30 – 05:36
Definitely Define a purpose like who’s your target audience of using this chat bot.
Cao Qing
05:36 – 06:02
What fat platforms you think they should be on should it be on your website on your mobile app, or should it be living on a off your native platforms like Google Alexa and potentially other smart speakers in the world, then determine the purpose of the box, then you can figure out the best way of doing it.
Cao Qing
06:02 – 06:03
Yeah.
Cao Qing
06:03 – 06:06
Well, I like a lot what you said there.
Ryan Lester
06:06 – 06:07
It’s almost like you’re
Ryan Lester
06:07 – 06:12
About like just like you would practive element process of you know, who is the person that’s going to be using it.
Ryan Lester
06:12 – 06:15
You know, what’s where do they live and work?
Ryan Lester
06:15 – 06:17
What’s the pain you’re trying to solve for them.
Ryan Lester
06:17 – 06:33
So just as if you’re designing a product, you know living in the world of software development, you know, let’s think about who’s the user is where they work what information are trying to access and then what’s the right way to then build a solution to service that customer whether it’s built in house or build with with a partner.
Ryan Lester
06:33 – 06:35
So, I mean, it’s a really great approach.
Cao Qing
06:36 – 06:44
Absolutely, and you see a lot of good examples other people uses one of the bank’s I really like they’re smart.
Cao Qing
06:44 – 06:46
They’re aii.
Cao Qing
06:46 – 06:53
So if you’re asking a question, it will actually guide you to the page in that mobile app.
Cao Qing
06:53 – 07:06
So if you have a really complex app with a lot of things embedded into that app, and this AI or chat Bots can help with faster navigation.
Cao Qing
07:06 – 07:10
So that’s that’s a very good case study, right?
Cao Qing
07:10 – 07:16
So that’s a method or you just simply want QA or you want to sell products.
Cao Qing
07:16 – 07:21
So there are many many many possibilities with the chat bot.
Cao Qing
07:21 – 07:36
There’s a couple things I like that you said there is one of those back to your original point of it’s not a one size fits all so you want to be thoughtful of you know, what’s the core problem you’re trying to solve and then what’s the right technology or platform to do it with and then to I like that.
Ryan Lester
07:36 – 07:40
I think often people think about Chad pots just as that conversational window.
Ryan Lester
07:40 – 07:44
So it’s hey, it’s just some text interaction back and forth.
Ryan Lester
07:44 – 08:19
It’s simple question and answer but in reality you can use your Bot to as a way to guide people to a lot of content content in your mobile app content on your website even things like PDFs or other digital content and it may be that is Services it up in the window or it may be that just redirects the person to that page and it’s almost like if you think you’re sitting with someone trying to find information on a website the Bop becomes that little guide kind of making the breadth of information you have on your website much more accessible than someone trying to navigate it to it themselves.
Cao Qing
08:21 – 08:42
Absolutely, and I would add in that you you can build in apis to your chat bot that can actually fulfill needs or request by the customers like ours you can do tracking if you have UPS my choice, which is a service that you can get notified when you have packages coming.
Cao Qing
08:42 – 09:02
So if you have a cat like that you can go to Google or Alexa and ask them to change your to deliver your packages to a UPS Store when you’re away like you can do those in the chat bot controlled by voice but supported by apis on the back end.
Cao Qing
09:02 – 09:03
Yeah, that’s great.
Ryan Lester
09:03 – 09:20
That’s a great example and it so my guess is though you guys didn’t do it on day one that was probably something that came later but you leverage that kind of core capability in the extended you reach into voice and then into more completion of tasks into assistance Beyond just information sharing.
Ryan Lester
09:21 – 09:21
Is that correct?
Ryan Lester
09:21 – 09:23
Yeah, absolutely.
Cao Qing
09:24 – 09:26
You also listen to the customers, right?
Cao Qing
09:26 – 09:29
We launched a version one on Alexa.
Cao Qing
09:29 – 09:46
It wasn’t a people were asking I can’t do anything about it because voice has a limitation our tracking number starts with Wednesday and the like so or Google would confuse like especially Alexa confuses onesie with a baby’s onesie onesies.
Cao Qing
09:46 – 09:50
So it never picked up a tracking number correctly.
Cao Qing
09:50 – 09:56
So people can really track with tracking number Alexa, but when we first launched it it was not very successful.
Cao Qing:
09:57 – 10:07
So we’re like What do we do when people in the you know responses in the reviews that why don’t you add in my choice, which is the service I just mentioned earlier.
Cao Qing:
10:07 – 10:12
We’re like, oh, yeah, then you can track without tracking number as long as you locked in.
Cao Qing:
10:13 – 10:16
So as soon as people start mentioning those were like, let’s do it again.
Cao Qing:
10:16 – 10:17
Let’s do it again.
Cao Qing:
10:17 – 10:20
So listen to your customer and
Cao Qing:
10:21 – 10:30
Realize there are limitations still nowadays with these platforms and really built the futures of cording lie.
Ryan Lester
10:30 – 10:32
So that’s terrific.
Ryan Lester
10:32 – 10:34
Yeah, that’s a good that’s a great story.
Ryan Lester
10:34 – 10:45
It’s also really great to in the in that I really like thinking about two of you want to make the chat box perience, you know helpful and useful.
Ryan Lester
10:45 – 10:47
It’s a new for many people.
Ryan Lester
10:47 – 11:08
It’s a new channel and you don’t know how people are going to use it and you know, you think you do and then you put it out there and I think the key is just as you’re saying, you know, you put it out and you let it learn and you you stay close to it, you’re monitoring it and you’re getting real-time customer feedback and the power is because it’s digital just like your website you can see what users are doing.
Ryan Lester
11:08 – 11:20
You can track their behavior, you know, you can get real time feedback to then course correct as needed and I think what’s really neat is if for you guys you might have said hey that you know that use case or
Ryan Lester
11:20 – 11:32
Integrating it with your existing, you know program wasn’t something you would have thought about it the gates but then based on customer usage and demand that became an obvious use case and something that’s what’s quite powerful about.
Ryan Lester
11:32 – 11:42
This is you can put it out there and then quickly users will figure out where it works well and where it doesn’t then you can start to adapt and and drive new use cases quite rapidly.
Ryan Lester
11:43 – 11:48
Yeah, that’s great and charge you another point.
Ryan Lester
11:49 – 11:50
No, I’m good.
Ryan Lester
11:50 – 11:57
Okay as so they I want to talk about so we talked a little about Knowledge Management and the platform.
Ryan Lester
11:57 – 12:11
I think another thing that often gets missed in some of this and I think there’s a bias certainly in the u.s. About this but you know, as you scale and you look at other languages, I think this is something that people maybe aren’t thoughtful of out of the gate.
Ryan Lester
12:11 – 12:17
So I like to explore you know, the work you’ve done is where did you start with languages?
Ryan Lester
12:17 – 12:18
And then where did you go?
Ryan Lester
12:18 – 12:27
And what were some of the challenges you ran into as you start to look at languages Beyond just let’s say, you know English or North American Central languages like English French and Spanish.
Ryan Lester
12:28 – 12:32
Right, so we are still on our way.
Cao Qing:
12:32 – 12:34
We’re still in the evaluation phase.
Cao Qing:
12:34 – 13:01
This is much more complex than a it is it sounds like I would say so you have several problems here one is scalability when you talk about scalability you have scalability 4 plus 4 and then you have skill abilities for language and then different countries depending on if you have apis there are limitations in four countries.
Cao Qing:
13:01 – 13:04
So let’s start with platforms.
Cao Qing:
13:04 – 13:14
So one of the challenges we see is that if you’re building a bot potentially you would build it multiple times for different platforms.
Cao Qing:
13:14 – 13:20
Just depending on the different requirements or limitations of each platform.
Cao Qing:
13:20 – 13:28
So after a while if you keep watching if you have a very ambitious goal you want to be everywhere then you would write.
Cao Qing:
13:28 – 13:29
To the problem.
Cao Qing:
13:29 – 13:43
How big is your team has to be two men ten all these platforms all all these different answers for each platform because some of them won’t let you send a link some of them will so yes.
Cao Qing:
13:43 – 13:47
So that’s a struggle number one.
Cao Qing:
13:47 – 13:48
It’s Justin.
Ryan Lester
13:48 – 13:53
Yeah, just a quick on this to we’re certainly feeling this also and you look like messaging as like platform.
Ryan Lester
13:53 – 14:06
So like you know, what what’s app will let you do versus WeChat versus Facebook Messenger like there are some commonalities in API, but there are also some dramatic differences just to your point like the rich typically.
Ryan Lester
14:06 – 14:09
Can you send a link in your not can you send an image file?
Ryan Lester
14:09 – 14:11
Is there a foul limit sighs?
Ryan Lester
14:11 – 14:21
So I think you’re spot on that as you integrate with more and more of these platforms or Partners the complexity quickly become you can just have whole team’s just working on keeping up with your Integrations.
Ryan Lester
14:22 – 14:27
Absolutely, and then not just messaging you have it on mobile app different screen size.
Cao Qing:
14:27 – 14:30
I mobile app apps and screens like laptop screens.
Cao Qing:
14:30 – 14:34
Now you have voice not just voice.
Cao Qing:
14:34 – 14:34
Now.
Cao Qing:
14:34 – 14:37
We have voice with screen right?
Cao Qing:
14:37 – 14:39
All these you have to take into consideration.
Cao Qing:
14:39 – 14:42
What is your eye you acts and all that.
Cao Qing:
14:42 – 14:45
So that’s a that’s definitely a big challenge.
Cao Qing:
14:46 – 14:49
So do you want to move on to next one?
Cao Qing:
14:49 – 14:50
Yeah.
Cao Qing:
14:50 – 14:50
Yeah.
Cao Qing:
14:50 – 14:50
Yeah.
Ryan Lester
14:50 – 14:52
Sure.
Ryan Lester
14:52 – 14:52
Yeah.
Ryan Lester
14:52 – 15:09
So language one is is like as you just said like I would point you translate so a while back when the translations weren’t the those type of software’s where Solutions weren’t as accurate.
Cao Qing:
15:09 – 15:11
You can’t really rely on it.
Cao Qing:
15:11 – 15:15
But nowadays those technology or so much more mature.
Cao Qing:
15:15 – 15:22
You can actually leverage the translation tools out there to translate the customer in
Cao Qing:
15:22 – 15:28
It however as a company you want to control the output, right?
Cao Qing:
15:28 – 15:37
So then you’re talking about managing all the content you’re putting out in how their modeling language you want to support.
Cao Qing:
15:37 – 15:40
So that’s a big translation task as well.
Ryan Lester
15:42 – 15:52
And so it sounds like for you guys because you know when we talked with folks here, there are a number of approaches and I agree with you that the quality of the translation services have gotten much better.
Ryan Lester
15:52 – 16:12
We work with Geo fluent in their line Bridge of the van, you know on the Fly translation service where you can take a customer input and translate it with really high accuracy leveraging AI to then translate it to understand intent, but then I hear your point about then what do you want to do about serving up your content?
Ryan Lester
16:12 – 16:13
So it’s interesting.
Ryan Lester
16:13 – 16:26
So you chose to go on the road of building your content in each language or at least the languages you want to support you’re not building like a centralized knowledge base and then Translating that to than what you send to the customer.
Ryan Lester
16:26 – 16:29
So you’re building out each knowledge base in each language.
Ryan Lester
16:29 – 16:34
So I’m not a hundred percent sure on that yet because we haven’t got to that part.
Cao Qing:
16:34 – 16:35
I got you.
Ryan Lester
16:35 – 16:42
Just depending on like UPS is in how our mobile app I think has been more than 20.
Cao Qing:
16:42 – 16:44
Languages yes, and we have our website right?
Cao Qing:
16:44 – 16:49
We have all of the a lot of the content translating those different languages.
Cao Qing:
16:49 – 16:54
So but if you have any suggestions, we would love to hear your opinion as well.
Cao Qing:
16:55 – 16:56
Yeah.
Ryan Lester
16:56 – 17:01
Yeah, so it’s interesting and I think you’re one point is a really good one of leveraging content.
Ryan Lester
17:01 – 17:02
You’ve already translated.
Ryan Lester
17:02 – 17:04
So certainly if you know, it’s saying, okay.
Ryan Lester
17:04 – 17:11
Well if we’re going to do simple things like maybe small simple text interactions, maybe we have a higher me.
Ryan Lester
17:11 – 17:25
We have a higher propensity or a higher ability to use an on-the-fly translation service because translating, you know, a hundred forty two hundred characters the service of gotten pretty good so you can say okay.
Ryan Lester
17:25 – 17:29
Well I’m going to build my conversational flows in one or two languages or five languages.
Ryan Lester
17:29 – 17:41
But but then I’m going to translate to other 20 languages I want to support but my my larger richer content things like articles or you know more
Ryan Lester
17:42 – 17:49
Policy document those I will translate because they’re going to live on my website and people are going to are going to be using those in those native languages.
Ryan Lester
17:49 – 18:08
So make sense to leverage those and I think this is kind of a case-by-case basis, but I think you bring up a really good point of, you know, you can do all this work to create a great experience and then the bot delivers like bad translated experience and that ruins it for the customer.
Ryan Lester
18:08 – 18:10
So like all the AI works great.
Ryan Lester
18:10 – 18:18
You did all the great work, but then you deliver kind of a, you know, a half translated final product and that really lets down the customer.
Ryan Lester
18:18 – 18:34
So I think it’s a it’s a really good point of you should be through really thoughtful and test this well of what’s the right way to manage your content as you want to scale to other languages, so it’s not just the AI part of it, but also the the content delivery part of it.
Ryan Lester
18:35 – 18:39
Absolutely great point and I do yeah gar.
Ryan Lester
18:39 – 18:40
Sorry got no.
Ryan Lester
18:40 – 18:41
No, go ahead.
Ryan Lester
18:41 – 18:57
Why hasn’t the last going to say that I really like that you brought up was also that there’s not only the challenges of platforms and as you know, when expand languages or expand reach and platforms, but also there are geographical concerns about data rules and data privacy.
Ryan Lester
18:57 – 19:08
And as you are moving to Europe with gdpr and as you move into mainland China the way you can access apis and access data really varies really varies quite a bit.
Ryan Lester
19:08 – 19:21
And so that’s another really thing to be thoughtful of as you can design something for North America or designer for Europe, but that architecture and structure may not be one that you can leverage in other geographies.
Ryan Lester
19:22 – 19:23
Yeah, absolutely.
Ryan Lester
19:23 – 19:26
So a lot to think about when building a child.
Cao Qing:
19:26 – 19:28
Yeah, absolutely.
Ryan Lester
19:28 – 19:31
So I want to see Japanese.
Ryan Lester
19:31 – 19:32
No, no God.
Ryan Lester
19:32 – 19:52
So another thing I was going to mention is there’s also a lot of talk related to what is can chat bot be truly a I were can you build let’s say machine learning models to Trend the bots so that there’s less people working on the dialogue clothes.
Cao Qing:
19:52 – 20:06
So that’s another area that we really want to exploring is how do you leverage unstructured or 30 data to Trend a chatbot that to expand the knowledge base?
Ryan Lester
20:06 – 20:08
That’s great.
Ryan Lester
20:08 – 20:12
And and so this I have heard this is becoming a more interesting topic.
Ryan Lester
20:12 – 20:13
Now.
Ryan Lester
20:13 – 20:28
Is this something that you guys think you’d build your own competency around or you feel like hey, we have some data and now we’re going to go to a partner or go to a technology platform and give them that data to help us understand these insights or do you think are going to try to build some of this competency internally?
Ryan Lester
20:29 – 20:40
Currently we are not considering internally, but this is definitely I don’t I don’t know if I’m giving out secrets.
Ryan Lester
20:40 – 20:53
So what I’ll ask the question so we’ll go back to where I asked the question of like is this something you’re thinking about doing internally and you can say something more like, you know, we’re looking at a variety of options or we also cut this action.
Ryan Lester
20:53 – 20:59
If you want you want to cut it or you want to give them more generic answer kind of will be probably better.
Cao Qing:
20:59 – 21:00
I don’t think so.
Ryan Lester
21:00 – 21:05
I’ll go back to saying so what will restart now.
Ryan Lester
21:05 – 21:14
So yeah, I think that’s a really neat next evolution of the technology where it’s not just humans kind of inputting insights into the bot.
Ryan Lester
21:14 – 21:19
But rather, you know, how does the bot start to identify where there’s gaps where there’s opportunities.
Ryan Lester
21:19 – 21:29
It becomes kind of its own teacher just like if you had a real agent working, you know as they started on the job and they did the job longer they’d start to identify New Opportunities new
Ryan Lester
21:29 – 21:33
Is to work so I certainly think it’s an interesting next phase of this technology.
Ryan Lester
21:34 – 21:35
Absolutely.
Ryan Lester
21:36 – 21:41
So you private a lot of really great insights and I know we’re running running out of time quickly.
Ryan Lester
21:41 – 22:00
But you know, you talk to Great really well about you know, understand your core use case you talked a lot about they’ll be thoughtful in the technology you choose in you know, how you want to build up your knowledge and manage that knowledge then how that scales to new platforms and how the skills globally.
Ryan Lester
22:00 – 22:15
So I really have enjoyed the dialogue today, I guess in closing, you know to put you on the spot a little bit is are there one or two things that if someone’s going to just starting out this journey or maybe they’ve tried and failed if there were one or two things that you would say.
Ryan Lester
22:15 – 22:22
Hey, here’s some right way to approach or some advice you have for folks that are kind of entering the world of chat Bots.
Cao Qing:
22:25 – 22:26
I’m trying to think.
Ryan Lester
22:29 – 22:30
Hmm.
Cao Qing:
22:30 – 22:31
What do you suggest?
Ryan Lester
22:32 – 22:40
So so I would say well so I think a couple things that and by the way we can cut and then you can give a one answer.
Ryan Lester
22:40 – 22:41
Okay.
Ryan Lester
22:41 – 22:57
So one thing I think you’ve done a really good job is to maybe take a good answer for you be like don’t take a one-size-fits-all approach in a sense of you should test and learn and and like what might work in one place may not work elsewhere.
Ryan Lester
22:57 – 23:00
And so, you know, really try to try to try to take it.
Ryan Lester
23:00 – 23:11
I mean even like your lean Six Sigma background, it’s like taking operational approach and you know build your process test your process learn from it and then, you know figure out how you can improve it.
Ryan Lester
23:11 – 23:13
So maybe something like that would be good.
Ryan Lester
23:14 – 23:16
Okay, so I said fine.
Ryan Lester
23:16 – 23:17
Okay, cool.
Ryan Lester
23:18 – 23:18
Yes.
Ryan Lester
23:18 – 23:18
I’m right.
Ryan Lester
23:18 – 23:20
I was just thinking about it.
Cao Qing:
23:20 – 23:20
Okay.
Cao Qing:
23:21 – 23:22
Whenever you’re ready.
Ryan Lester
23:24 – 23:25
I’m ready.
Cao Qing:
23:25 – 23:26
Go ahead.
Cao Qing:
23:26 – 23:30
So I would say that building a chatbot is definitely fun.
Cao Qing:
23:30 – 23:32
It is in Trend right now.
Cao Qing:
23:32 – 24:04
So I would say have fun during the process but definitely test them learn there are many solutions out there tests and figure out what pizzas your tests tests and figure out what fits your situation the best and then don’t be afraid of failing because they’re there are a lot of thoughts that are failing out there right now, but what happens is you need to learn from it and make a better version the next try next go around so that would be my suggestion.
Ryan Lester
24:04 – 24:06
I think that’s great.
Ryan Lester
24:06 – 24:11
I think that you know, sometimes it can seem a bit daunting and there is a complexity to it.
Ryan Lester
24:11 – 24:24
But really in reality you can deliver a much better experience and you shown this you’ve given your your customers much better access to the way they track and manage their deliveries and you’re doing it through
Ryan Lester
24:24 – 24:30
Global platform that can, you know work across channels can engage people and new channels.
Ryan Lester
24:30 – 24:31
So I think you’re spot on that.
Ryan Lester
24:31 – 24:42
It’s really an opportunity to get something out there, see how users use it adapt and then continue to grow the capabilities in the breadth of where chatbots can change your business.
Ryan Lester
24:42 – 24:47
So, I think that’s great advice what will change thank you so much.
Ryan Lester
24:47 – 24:58
I really have enjoyed the time and looking forward to our next conversation sometime in the future and best of luck on your continued journey and making chatbots a great experience.
Ryan Lester
24:58 – 25:00
Thank you very much Ryan.
Cao Qing:
25:00 – 25:01
Really?
Cao Qing:
25:01 – 25:02
Appreciate it.
Cao Qing:
25:02 – 25:03
Yeah, my pleasure.
Ryan Lester
25:03 – 25:04
We’ll talk again soon.
Ryan Lester
25:04 – 25:05
Thanks.
Ryan Lester
25:05 – 25:05
All right.
Ryan Lester
25:05 – 25:08
Talk later, bye-bye.


Discuss / Read Comments

Leave a Reply

Leave a Reply

Explore our other AI Insights or Recent Posts.

Subscribe to the Bold360 Blog

Submit your email to get updates right to your inbox.