Events Add an event Speakers Talks Collections Sign out
 
Duration 44:55
16+
Play
Video

Beyond PoCs: Choosing and Operationalizing a Conversational AI platform for Enterprise Scale | Rasa

Alex Weidauer
Co-Founder & CEO at Rasa
+ 4 speakers
  • Video
  • Table of contents
  • Video
Rasa Summit 2021
February 11, 2021, Online, USA
Rasa Summit 2021
Request Q&A
Video
Beyond PoCs: Choosing and Operationalizing a Conversational AI platform for Enterprise Scale | Rasa
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Add to favorites
291
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About speakers

Alex Weidauer
Co-Founder & CEO at Rasa
Casey Phillips
Sr Product Manager, Conversational AI & Messaging at Intuit
Vineet Malhotra
Partner, Digital Ventures and Alpha Labs at Mercer
Dennis Yang
Lead Product Manager, Conversational AI at Dashbot
Sweta Patel
Sr. Director Product Transformation at Juniper Networks

Accomplished Product Manager in the Conversational AI / Chatbot space, with over three years of prior experience in software engineering, in addition to an MBA degree. The ability to effectively monetize and utilize AI & ML in products is one of my greatest strengths. I’ve achieved notable success managing the product roadmap for a variety of platforms including mobile apps, cloud-based solutions, enterprise-level software, and conversational UIs (chatbots and voice assistants). Recently, I've launched successful and groundbreaking Conversational AI experiences in prominent products including TurboTax, QuickBooks, and ADP's HR Portal, all of which have delighted users and exceeded business goals.

View the profile

Digital experience spanning over 20 years across financial services, telecommunications and education. Specialties: Digital Strategy, User Experience, Data-Driven Insights, Behavioral Science, Social, Operations, Branding & Marketing

View the profile

Dennis Yang is co-founder and CPO of Dashbot, an analytics platform for bots. Prior to Dashbot, he co-founded Bureau of Trade, a vintage marketplace (acquired by eBay), co-founded Techdirt, the popular technology blog, and was the ninth employee of both Infochimps (acquired by CSC) and mySimon (acquired by CNET). He earned his engineering degree from Cornell University and lives in San Francisco with his wife Tricia, son Jax, and dog Bear, who introduced them.

View the profile

About the talk

Many enterprises have run dozens of proof of concept for text- and voice-based use cases in conversational AI - from internal to customer-facing use cases. What did they learn? How do they think about choosing the right platform for their enterprise needs? What team and processes do they put in place to fully operationalize conversational AI?

Moderated by Rasa CEO and Co-founder, Alex Weidauer

Panelists:

Sweta Patel: Sr. Director Product Transformation at Juniper Networks

Vineet Malhotra: Partner, Global Digital Alpha Labs Leader at Mercer

Dennis Yang: Lead Product Manager for Conversational AI at Chime

Casey Phillips: Sr Product Manager, AI-Driven Assisted Experiences at Intuit

- Learn more about Rasa: [https://rasa.com​](https://www.youtube.com/redirect?even...​)

- Rasa documentation: [http://rasa.com/docs​](https://www.youtube.com/redirect?even...​)

- Join the Rasa Community: [https://forum.rasa.com​](https://www.youtube.com/redirect?even...​)

- Twitter: [https://twitter.com/Rasa_HQ​](https://www.youtube.com/redirect?even...​)

- Facebook: [https://www.facebook.com/RasaHQ​](https://www.youtube.com/redirect?even...​)

- Linkedin: [https://www.linkedin.com/company/rasa​](https://www.youtube.com/redirect?even...​)

#conversationalAI​ #enterpriseai​ #aichatbot​

Share

Welcome everyone. And I'm really excited for the next like 40 minutes, the panel Beyond POC as a proof-of-concept, choosing an operation lies in your conversation, AI platform for enterprise-scale. My name is Alex and the co-founder and CEO. MB overhead Raza. And I'm super excited as a panel kit with like leading experts in the field that have seen a lot of enterprise-scale. And before we dive into this discussion and I just kind of wanted to quickly introduce the topic a little bit. And so when we talked about like proof-of-concept and I think

everyone here, I'm in the audience is aware that basically almost all companies in the world have run, certain proof of concept in the pilot project in conversational AI. Like to take space, boys, based, of course, all started with like Facebook messenger and opening up their platform in 2016 and all the other developments that have happened. Since then, write interview. Look at conversationally. I am the assistant. It's like the next kind of wave of interfaces and your question read. It becomes like, how are they build, right? I mean, people know how web apps and mobile apps, iOS

appealed. So, this is kind of an idea of like, what platforms you can use to build them but also like Frosty's how to operationalize them and really applications. But I think that question is a bit unclear still when it comes to conversational Ai. And so, when we talk about a conversation, I really mean is like the foundational and technology for developing the chat button, The Voice Assistant, but I think it's actually pretty good definition. And then the question becomes like, you know, what does it mean in practice, right? Because everyone wants to be a pest

problem these days. If you mark yourself as a platform, that's the safe way to get there, and but Any more practical like with a few people in the Enterprise have seen being useful for them. And so, when you think about that, and the way I look at the conversation, they really like your business and your customers, right? So on the one inside that you have customers, that connect you on a lot of different channels to look different places and then you have your company and like the business. Shaking in your company. Write, like,

I'm basically a conversation in the middle and it can protect space or boy, space avoids, meaning for examples of telephony. And and then it really becomes like this question around like how can you automate conversations with your customers and push down to the like let's save mobile app. For telegram is another conversation and basically get a response from something in your company, right? And it might be like a knowledge base, it might be some robotic process automation. It might be actually a human agent as well if you want a route to that person M

into when we talked about enterprise-scale this is just it's not just like this like two channels that I just described as much as the mobile haven't had a gram, but it's like maybe at Facebook Messenger, maybe it's Android. Maybe it's some other like channels that use maybe they give call Santa right now. But then it's also not just like this one system of a couple of systems that you would plug into in a smaller company but just like these billions of systems and in this is like an example of what we see out there in the wild and and it's not just limited

to that. Because in the end, if you are an electric company, you've probably like certain quality assurance processes that you need to follow certain approval processes. And you might like have complete the weather like in human languages, right? Like natural languages. You might be in different Geographic markets in the world and you probably have a larger team that is working on that. So you need to figure out like how you can basically like dissect your assistant into smaller pieces and you're certain requirements around deployment and you just have increased traffic right through the

millions of conversations and handling. This is not only a question of deployment system. Any other questions that are related to that? Excellent addition to that. Coming back to this team and what we also see happening in nice companies that. Is that you basically from multidisciplinary teams and to really build and run these texts and voice assistant. So I would have maybe like the product owner and conversation designer data, scientist, developers content, owners, and so they all work together the same way as those types of teams also, work together to maybe ship a mobile

app, web application to relieve his assistant. And so with the super quick introduction and I want to get going into our discussion and I'm really excited for all the folks that have been able to join here today and all of them have seen enterprise-scale other than conversation and to keep things off. I would love to have for all of you to do this quick and did you see this house? And we get going, Shrimp salad. My name is 2400 singly conversational. AI is is at the Kauffman and experiences that we not only have been experimenting with but to Alex

looking at that price as well. Thank you, Dennis. Everyone thinks they're helping out like Sam. I'm going to Shanghai in the lead project manager for conversationally. I chimed company that provides banking services in the Premier League for the Michiana. We believe that everyone deserves kind of financial peace of mind. So that's time. I prior to this, I've been a part of going to five different startup startup was that we we provide I analytics, and optimization board conversation face is

like a pot. So that's that's where I got the bulk of my experience that in this industry and I'm excited to chat about what that means for Enterprises. Go flex, tennis sweetheart, are you in conversational AI? I spilled a few large-scale. A conversational. AI products over the past seven years. One of them is Erica from Bank of America. And now the one is Oracle virtual assistant from Oracle and its price light and currently, I'm driving park, has formation at Juniper Networks and the Vikings and

startups in the space as well. Cool. Thank you. I'm in MKC. Thanks for inviting us for going on 5 years now from management perspective. So I've had the opportunity to launch bill for larger companies and now my focus is on bringing about the bring you out the AI and ml modeling that's on the Ocean between The Choppa and the life expert experience. So pretty excited for the panel. Call me awesome. I think we can just dive right into it. And I mean, I mentioned earlier that I was quoting has been one of many

that they've done in this field and I think if you talk to them, they will tell you the Thera about, like, 2,000 like vendors of conversational, AI worldwide, right? Like it's a huge amount. And so I think it would be really interesting for the audience and to get started with today is really like from your experience like what different approaches have you seen in the Enterprise and to build text or by space assistant or both and and maybe also talked a little bit about like the traitors and they do seem like one size fit all the way to the end. And I think this is something that would be

super dressing to get a little baby bum. Yeah, I'd like to check I kick it off with some of my thoughts are so no from me. I'm working for fuselage of companies which have also stepped on one of the most heated debates that I have the actual to Bayonne. She have the more box or like your body. I think there's like a large come get this into a priest or a large companies that call. We, I mean, there's so many different use cases for Chiapas new, you talking with different products in a completely different subsets of these

are questions that office that office and just about to say to you, you try to have fewer box that archive more complex and if there's a lot more rules and Islam where logic in there, just a lot a lot easier to ask a question, or do you try to more so fragmented and have more Bots are a little bit more like to find a little bit and you can probably move in You have a little more jiloty, but then this kind of comes to the, then it kind of comes to the conversation about just doing a consistent experience at. So, if you think about the easiest

way to have like a consistent conversational, AI experience on a large enterprise-level, will be to have fewer box but then again like that takes more time. Like I feel like if you have like your boss again, more complex you can move quite as quicksand and they're on the other side of the fence, you have just constantly more box in the smaller. You can move a lot quicker, but then you really have to get things down down from the operations and I keep on processes, they going to keep all these different teams, all these different campsites in check so that I still have conversation with

QuickBooks or TurboTax, or maybe you're in a different countries are still getting that cop consistent experience. So that I find like that for me is always been as one of the most contentious topics. Enterprise chatbot. Dammit, question, can I tell when when it's actually a company takes these conversational AI experiences and conditions than from typically like The Innovation team over to the kind of the business unit. The key difference here is bad but the goal behind creating the conversation Eliezer chatbots shifts from make this

because it's new and do this for certain business purpose. And I think that that's what I picked up at Casey's getting all sides which is now the decision that determines whether or not you make multiple Tapout. You're not really boils down to the first decision point that I think you need to make when you transition into the into the real business unit which is why are you doing this? Why you making these conversation on your face? Is that all you know, it's sort of looks back at you know the web days. The mobile day is going to be used to be. You have a mobile team and a web team. And

now it's centrally. These are essentially Solutions fax that are addressing kind of real business problems. And that to me is like the single biggest shift when you shift from in the Enterprise, is it? No longer becomes and I were doing it because it's new it. So, we're doing this because this solution is the best solution for the problem. And I and I actually find it a little bit odd that, you know, as a product manager I guess my title is heading the conversation a team and that's a solution for a solution for the product manager. But I think that that role and of my my my

goal, my mission is to essentially like be the consistent platform with which the rest of the park managers around the organization or decline kind of conversational AI Solutions. I'm in the same way that no centralized Design Systems teams are looking to make sure that the consensus consistency of Design Systems are deployed across the company. My mission really is to make sure that The consistent approach framework and invoice deployed across the rest of the Enterprise as you develop more conversational AI. So definitely like Casey I I hear your pain. When you know, do we make one assistant

from several assistants huge, huge issue but I think that's the foundation upon, which I can, I can send that kind of message across my company. Is that here, you're the best practices when deploying Enterprise systems across our company and make sure that we did at in an efficient and consistently great point. And if if I may come in, you know, some of these strategic objectives that are identified in enabling conversational AI interface has, you know, who are critical to identify enter and do what he got to go ahead ahead of time or

early on before we go into the solution space, right? Because the solutions play so it looks like at least I classify it into three different areas. You can take a Integrated Solutions into existing digital assistants. Write the existing facilities of the iot box of access or you can have in hybrid of, you know, utilize open to an API stand and then to rebuild the business workflows and processes inside that or then you go completely customized environment and completely customized platform depending on the

complexities that that are associated with these fundamental to finding a path on where the road back for the products. And also from a customer service, point of view the design Integrations, do you use cases of word? The value addition, and an embassy X identification. All of that sort of comes in from that the identification of what the Enterprise strategy is. Speaking about trade-offs on scale, I do. I do agree with you guys on, you know, the multiple box. Purses. You know, one box that, that sort of has multiple skills affect offered and

I could have seen a boat. I've actually, you know, I had a Erika. For example, if there is a great implementations which is pretty much a banking experts and you know, she she actually walks the user so many really good Financial. He was cases and Uniform Florence. And proactively help them manage, right? Where as some of the other implementations that I've seen have been more specific end and the leaner in in their use cases and not as complex. But yet, they are simpler for from aux point of you, for, for providing the customer service setting, it depends

from a scale point of view and to Alex's, the previous slide. One of the things that you mentioned, Alex, You know the channels increase access key and increases as the scale of customer. Interaction increases the variety of the term, recall, omni-channel presence, right? That increases. And that also brings its own set of interaction challenges and also design scale for the conversational AI. But I think some of these are very, very fundamental again. It was back into what's the original strategic objective is for the Enterprise

iOS at the same time like this that I'm going to try not to repeat. What was just said, one of the things that we've benefited from is that I am You look at it from a home life into work life. Respective. Thanks to the whole life experience of our colleagues have a company that using a Google assistant. So they're reasonably familiar. I'm also in the last 12 plus months, in Sam's of automating person, that has always been his expectation is Lena. Hey, what can I do? An Alexa in the workplace kind of thing. That sucks is

there's been some sort of education has already happened because people expect it all to the same detecting scale. May rest assured the technology to scale but can be experienced kill. That's the challenge where I can be created his own experience and skill as well. And how often do we design experiences to Scout? And then when it comes to getting the experience with conversational, AI the approach that I've taken his sister to say, to various business teams of product teams is if you want to Put a chartplotter virtual assistant experience on your

product, awareness, and action like that, how you going to use it for particular set of use cases, so they can access all of that. Cool functionality. And then are you going to bring actions to the user with the charcoal? Are you going to take them to the action of applications sub? I'm trying to take off, a behavioral psychology be off this because the technology has been shared so much and we've been using one another for a while. Now, we know that it's it's not even about the technology so much

anymore. It's a bad really applying it and then creating experiences that can scale as well. I think it's a really great find actually about the experience of scaling. This is the tech, I think so too. And also angle that we, at least see all day in the wild and being sometimes, not only a challenge of conversational AI itself, right? I think like a lot of the, you know, back in connectors, Etc, and large companies are still sometimes not a new place where you can actually create those experienced. By the minimum was joking. And when, you know,

traveling is visiting customers in real. Life was still a thing before covid. I was trying to connect my MacBook just to, like a present her in the room and it usually never work, right? And I mean, it's a small example of a degrading things together in a company, but at least, I think that's the challenge. Should we all see all the time? Yeah. But, you know, that gets in addition to the technology stealing is, I really think the transition between Innovation and Enterprise. It's also making sure that the organization in the process, you still behind it as

well. I mean, I think typically I, you know, I see one person making a chatbot kind of went when it when it's an innovation, right? So they're making the training day at other designer, be intense. There's any conversation to doing everything and that's really easy because it just one person. But then how do you actually make that scale when you're trying to do it with me? There's too much work for one person. So then, how do you even start splitting up and creating different roles? I'm different at different in different processes. And also then, tying in like knowledge management, all

across the organization. That's what really think enterprise-scale means. Like, how do you, how do you create a conversation on your faces? When you when you shift from a team of one? Who knows everything into a team of ten? Twenty Thirty, beat me up. Plus people that are I have kind of stay colors all around the organization and they're coordinating information and process is all across. And how do you keep coming from that point into, very key areas that right up, what is Knowledge Management? As he mentioned, really do application

of data sources, and specifically, tied around the other picture of use cases, that we went to deliver to the user's, right? And then the other is the process in which is a variety of prostitutes from different departments. You know, product has its own life cycle ID has its own operations, it will show you how they deal with and stuff and do things. One of the one of the critical things conversational, AI technology print is it demands? Aggregation of multifaceted little prophecies add knowledge base in a way that no other system. So far has actually demanded, right?

Because this is a unified interaction and it's if it's one entity that interacts with the customer, It's not a mobile app with multiple menu that the customer can click on it. It sort of works of different different processes in the back and just doesn't work like that. But I think that is that is a huge challenge in Enterprise in Oak. Brook from a scale point of view, and implementation point of view, that two teams have to work on the back end and I think majority of the time is actually spend like, we getting used to things, we need the data and the properties and the Devore

integrated. This is the higher, the quality of the conversation. We are experienced in, in my opinion. They should get points. And maybe something with conversation is is, is is the conversational AI becoming relief. Petition is a disability in this binary, like, spitting up a check, but putting it out there, really binary, FAQ set of stuff. Got me saved up, you know, uses your clients expected to be less binary. They want to have a conversation. AI is a difference to the trade conversations with a different direction than expected management

discuss the risk with these kinds of experiences. That is brutal and you get it wrong and it's not a kiss and not coming back until half is deleted. Full set of clamps to the needle to tragically Center, figure out how you make them less binary, how you train your models to be. Most of testicular, still learning and getting better at it, but not where we need to be. But it's something we got to be really conscious of his well, it's a real risk. Casey, maybe, if you, if you think about your team, it into it, like, like how does it look now, like, who's part of it, and also maybe how it has

evolved over the last couple of years? Yeah, I mean, there's there's a lot of teams involved right now in regards to just everything, we're doing a conversational AI at at into it it's really grown a lot to me. That I would say right now, especially is fans, at least like three different words. In several different teams are part owners. You got your conversation designer, you've also got like the people that can't even talk anymore. So I kept talking like The Innovation side and then I got

the people that stay on. The beach is definitely a lot. A lot of players. I really like the point that dentist properly, just tell her to send that as you, try to release Gail these experiences. And in, in an Enterprise environment, Just take the focus. I really haven't been in the house, like they're the people and processes just alignment is so important because it can be, it can become very easy for teams to call. I drift off a package out the conversation about like one chop off your having you do or if you were an organization,

you decided he didn't like you really need the alignment from those things. Because every single Choppa is going to be an extension of your brats of one chap. Like you just to check I have a certain expectation of of of all the chat box of your organization would have if you have multiple. I think I've seen have like a roadblock that's been hard for for Tuesday, I get over this. We have all these these keys in the car, so sorry I spaced but then you try to get over

the hurt of globe. PlayStation is Wallace always been something that's been really challenging. If you're how do you support chat box in multiple multiple countries in the multiple languages and then it gets to you. I think you're from my standpoint it's always been what a horse things that they get in an Enterprise company. Once you try and go Global the amount of teams and then I'll play as you have just it exemplifies because you're talking like, you need to have people from each of those countries who are like experts on, you know, like, you know, how you're trying to chop off. Eunuchs

countries are really kind of strapped, right? Amount of people that you have working on your station from my biggest challenges. But that's actually a very critical point you're burning up. Because as, as the scale of the globalization of a checkbox or or conversation, they are really excited. Because we're going through the shift from Tropic, individual assistance as well. And I think the teams involved are getting deeper and deeper into complexities and and hence, the scale episode of, you know, where is that distributes from a Geo point of view as well? Because

you do, even if the business workflow is the same, you know, from one country to the other the language interaction brings its own set of variety on how that you know experience the natural language interaction experienced a sort of scale and I've seen where you know, we've had two buildings designed to you in a different country is just to cater to that language experience, right? Not just because it's one thing to have it at that then and have a conversation. Only I built for it. You think interacting in that language? But then it's a whole other thing to test and verify

and, and understand that the CX office and then bring that in an end to the algorithms for, of course, for that much energy. Also, I think that the complexity definitely go with multi-fold and I think the risk and compliance as the authority of mentioned it in a different context, it's still that extends as well, because you know, what's with the applicable here from a compliance, but if you wasn't exactly the same a snapshot in other countries, that scale occurs. So I think I I think it's if I'm beginning to wonder if that there are the unified teams that are Chio

specific that are being you know, expanded into geographies. And I think you're also, there's a big difference between shipping it once and having to continue to shift overtime, right? And improve. And because it's what you do, if you come out of POC stage and actually have a like product infected you with building, I think in many ways. And and then I think if you have to internationalize ones and you know there might be easy but then if you have to improve that overtime is. So I think it's really wet. The challenge

comes in. And I mean, we, we talked about this a little bit but like people in process is really like, like, how can you make sure that they are set up so they can continue to ship them quickly as well and improve the customer experience. And I mean all of the sudden not enough to like weeds and complexities of the Enterprise and what Ed scale really means, but I think we touched on a couple of months already and but if you think about it from your standpoint, Picking the right platform. That's

what really matters because it suddenly, you can't have it all. I think that's like a pretty clear inside here from the scheduling and so you have to make choices, right? And if you think about like the top three criterias, and for you to sing a conversation at that point, would that be a bond? What were what an Enterprise is? Kind of getting out of it? It really depends on Auntie classification of which platform, but I think some of the fundamental needs and then

to be number one, is really the business-to-business blow that the implementation, you know, assessment that we're trying to gain after the second is really the variety of data and then really the variability in data more so than the actual date of that. And how that the how was this term handles that right? OS through the peeling of the barrier. If you have to decide or even just me or if we use off the second one and I actually have four not not free. Because infrastructure I think it's a very critical aspect of this, right? And an octopus that do speaks to some of the risk and compliance

privacy and also scalability of how do you know, what's the volume of a cube is? You're trying to look for what's the complexity of the operations and stuffing and hands, you know which type of system? The rich. I preferred Solutions out there or homegrown, you know what type of Demands, the infrastructure demands these things have. And then the fourth one, that's more emerging now recently, at supposed to lock more before its analytics, write the ability to drive analytics and and and feed to Percy customisations off the digital assistants to me to cater to the Pacific Yew Cases

with customers are personalization eventually, I think that level of scale is also to be considered when it comes to picking the technology. Yeah, I mean that I think the way I look at choosing an Enterprise platform is really you start with your answer with your objective and then look at the entire process and to add as far as like the pieces that you need to create when creating a conversation. So if you think about, like all of the requirements that are there. Yes, at the very base. Level is the technology like good enough as a cute enough to the main said to do

your purpose. But then actually I think the larger pieces that I'm really looking at other things. Like and what does the deployment process look like when Minnesota played in Indian surprised? What is Version Control? Look like, what is experimentation? Look like, is there a good experimentation solution as in? This is integrated with analytics? Can I understand how certain unit dialog clothes are our stories perform when they're in production and can I really can break this down in a granular level of training, like, I'm going to do validation in an

automated fashion. When I'm improving my models and sending data. And how do I version to using attach these to the analytics that are being generated? When I had these systems in production of those? Are the questions that really? I'm starting to ask, you know, has as we are maturing as an organization that he's better than using these interfaces on Deck, how do I descale the improvements and the processes? And what is that look like? As I choose the platform has have have all of these factors that enterprises care about, you

know the burgeoning experimentation analytics, what it what what are the solutions look like? When when you're actually driving to Richmond by Mia? 110. Just a little stories like in a multiplayer. Is I think very fascinating meaning, like you? Look at her joyful. It is to design in figma where in It's amazing to see more multiple people working on the same exact document and that really, I think speaks to the strength that I think that product management team understanding that. This is how these things, as used in production. So, you know, I want to see what is it. Feel like when I

have a team of people working on these systems? Like, do we run into each other? How do we coordinate our versioning, how to record a who's working on why? Because there was the real types of day-to-day problems that that will be facing as we are and working on these interfaces. So it feels like, when you're maturing these platforms and solutions at the ability to experiment to have more than more than one person work on them, the ability to validate and test like you to all concerns that key to that the fact that these

are real applications being used for real things with by millions of people. Yeah and I I really like the point they just made Dennis to about just your how is the parking lot of people in your team to actually work and I'll just add to Kathy back. I think another thing that would be important to me I look at the clock or just from like the core apothicons. You know, how difficult is it to actually go in there? Beauchamp off the conversation? Cuz I'm sure everyone's had these friends arrive workers on some chop off platforms, where it feels like for someone to, actually go into Sky design

and develop very simple, basic Choppa conversation. They almost need to be an engineer, have an engineer back on. You really don't want that, that's going to make it very hard on an Enterprise, Al to truly scale because you may check there, so many people and players involved, and to think that everyone on the team is going to be able to think like an engineer. It is just talked to use when you're always going to want those people on a team that you don't come from an engineering background, but they truly understand like the customer called and they understand how they like actually

designed a good conversation. They don't have any in the engineering experience, you really need a platform that allows of all that core functionality to be very simple. So you can have more people more advanced capabilities, you shouldn't have to be having engine. They sell of engineering experience going in there and like, creating and authoring your most basic conversation clothes that you should just really have someone that I truly understands the customer Palmer, understand what's the experience of the customers needs and they can go in there and actually do it for myself. I

think that being able to touch, we built to scale and grow a lot faster, if you have to constantly have that go through, so on with that engineering background in there with a class. So those things up I would like to that commented in integration and interoperability, I really important because I'm, I don't want to have the conversation with. We don't want to create an army of child abuse cases of problems. You can say, is that I don't want to, I want to be introducing experiences. Where does applications already then organization

There's an opportunity to use a conversational, you iron a tie 2-2 to actually solve problems and quit making more efficient and and have this time Enterprise workflow orchestration that works with the chalkboard so I can go from one application to five others. We are thinking about it anymore. I'm just focusing on the task that needs to be done to this requires and call you level up in probability. Ease of integration where I'm focusing more on. And to enjoy, I need to spend 15 application sometime and this work flow castration

in the notification. For once before a long time, conversational AI, we have something really delightful to do that. I guess it's almost magical and waxing. Are you kind of go wild? Like I'm just probably going to five applications that I didn't realize it because I was having a conversation. I think that's the that's the promise of what we're talking about. Tony and I think you're all so it will probably won't be just like one platform that you don't just provide said all out of the box in the symptoms of UTI. Right? Because I think of anything Enterprises are extremely complex and and

that's probably something that a lot of people wish they would exist. But I guess in many ways doesn't exist until today and it reminds me of when I when he first got started in Fresno, I was visiting a call center in what agents that there's basically to like listen to what the customer tells them and electronics translated into the business logic of the company and then they would just look at like six or seven different screens which already like different, like programs the different tools basically and then input that data and they're right.

And I think that's what you just described across these six different applications that run somewhere in the back ends and I also think that's a big price for sure to be one. Only two minutes left and I feel like we just got started. So, I hope we can continue this conversation again at another conference and the dead. We also may be in a separate webinar or something like that, but maybe just a super quick to wrap it up and from each of you and like if you think about it like success ride like in and maybe just like throw it out there maybe

the teaser for the next panel to be organized, how would you measure success for an assistant? Yeah I think you're from my perspective to can I piggyback on what Dennis is as said before. I think it just leave everything like they used to your organization and what you're getting a lot of air pressure medication that I work on chatbots. With generally speaking the primary goal is Mekhi containment like reducing the contacts that I keep being stopped of questions that uses having for prevent them from high to go sign in the

context there. So they could at least in the chocolates, I work on like that's being kind of the predominant and whether you're measuring the rate of automation are just make sure I can tame it. When I'm right go or you met you that you're going to look at but he's just in a case for me historically Yeah. I think there are two aspects to to measuring the success, right? One you mentioned how much of the self-help and what does the frequency of use and and and even going from reputable used to in increasing our expansion in the youth

cases and adoption rights is is a good measure for success but I think internally from Enterprises as well as the operational efficiency that the thing brings, you know, and the integration of optimizations off the manual examples. You mentioned Alexis your contact center agents, looking at your 5:10 different screens and fabricating that information and putting it in a internally. That's also a measure of success. And, and that's really what's very closely tied to some of the ROI initiatives, you know.

After I guess for me, and like, that's got any and how can I physically improve the lives of my members died. So I'm I'm primarily focusing on satisfaction, and happiness and joy. Enjoy. I want, I might going to my customers and members when they interacted with my tripod to truly have like an amazing experience and if they want to talk to them as quickly as possible, if that's what, what if that's what's going to make that, like, raise up, make them resolve, their issues

quickly. So, that that's sort of exciting successes that, that is that I need to focus for. For me, is really around like the solution and not the problem that we're solving, which is he did we get Answered your question as quickly as possible and not necessarily around the fact that it was delivered by a conversation. I, I like so, it's sort of like the solution is Sal, Patel of an amazing solution should actually almost be secondary to the problems that we saw. Was there an end to me? That if if no more of that conversation

AI is used in the the tool chest of my fellow of product managers across my organization. And that's how I'm actually going to be judging My Success. Check my internal stakeholders to different parts of the business will come to me and say they want to use conversational, AI or the color different things. But they told me that he propositioned really thought through, I find it until class will come. And say that I would like to use this technology because

it's going to help my and customers get an answer quick up, get the job done without having to make a phone call and be on hold. So they come to the very clear value proposition. And so we focus on solving for that almost immediately experience about the ROI calculation. Pretty straightforward. It's it's it's people have talked through where the value business will be focused on. Maybe it wasn't the case four years ago. But I feel like there's been a level of

maturity that teams have achieved in the space. And I think that's a great point. And I think also a great finish the discussion and the other thing in the end, everyone can kind of understand like what an automated conversation could look like, right. I think it's, it's just really about making it happen. And I'm really happy that the four of you would be here today and she had your Lessons Learned. I think what else to take from that is like enterprise-scale complex and I think having a patent on that is also something super worthwhile happening, but there's so many more

things to go Mets. Go into debt. And I hope they like for the audience today, at least, like helps you to ask a couple of the important questions and and I think some of them might be answered in other sections, that brought us some of em and if not I think will also be here for questions afterwards. That today. I'm so thank you all and you have a good rest of the day. Great. Thank you. Thank you. I love you.

Cackle comments for the website

Buy this talk

Access to the talk “Beyond PoCs: Choosing and Operationalizing a Conversational AI platform for Enterprise Scale | Rasa”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “Rasa Summit 2021”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “Artificial Intelligence and Machine Learning”?

You might be interested in videos from this event

October 7 - 20, 2020
Online, Mountain View
19
5.41 K
google, googledevs, it, machinelearning, mlsummit, network, platform, tensorflow, tfx

Similar talks

Bruna Nayara Moreira Lima
Chatbot Developer at PicPay
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Buy this video

Video

Access to the talk “Beyond PoCs: Choosing and Operationalizing a Conversational AI platform for Enterprise Scale | Rasa”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
742 conferences
30258 speakers
11303 hours of content
Alex Weidauer
Casey Phillips
Vineet Malhotra
Dennis Yang
Sweta Patel