Digital Workforce - The rise of Enterprise AI (IPsoft)
Welcome. To our next episode, today's theme is focused, on AI and, digital labor and, joining me is air goon I said, alright very well okay, arrogant, Turkish name correct spot on all right wonderful so Eric on your with IP soft that's, correct, can you tell me a little bit about you and IP. Softest oh I sure. Can. Take quite a bit about it I've been here for 20 years and the company has been around for 20 years 20 years it, was seven of us that, started. Coming. Out of NYU and, starting. A little loft space in Soho some. 20 years ago. So. About the, late 90s, my, brother and. I our, CEO Jaden, and I, figure the pier is Gary Mike were at NYU and, Jaden. Was finishing his a dissertation. On, the. Term mystic finite state machines which. Is a very catchy. Topic, yeah. I. Blew the paper still collecting dust somewhere, in the library at NYU and. We thought I will chain thought about finding, a real world application. For. The research done on how, good machines. Apply. Decision. Trees to take on the role of common. Challenging. Human tasks okay. That sort of company was to see the orientation or the beginnings of AI this. You know the, beginnings of AI you, can I would blame mr. Alan Turing for that okay in the 1950s. And his paper. On you know he a proposed asked the question can machines think. This. Was however what I would say is probably, being as a practical application of, a oK we've talked about for a long time. There's. Been a lot of fiction around, it. Sadly. There's been very little reality, right, around especially when, we started robots. Taking over the world great. Movies movie magic but not, commensurate, with anybody that knows technology nowhere. Near to reality so. When we started leveraging, the term mystic finite state machines, to. Apply intelligent, machines to solve problems for humans in IT. You. Couldn't, sell AI back then because, nobody. Was talking about love it was buying right so, we actually started as a med service company, okay what that means is everybody. Was offshoring. And nearshoring, and. We said you know what we'll. Sell you that service, mister. We're gonna leverage digital. Big machine. Labor instead, of human labor can't, do that that, was our starting points okay ever, since then at some point we realized I talk too much. So, instead of just writing or the code anymore now I speak about the code we gotcha well this is why don't we start with the, very basic question because I think everyone, gets so confused with this what, is AI, artificial. Intelligence define. It for me first, and foremost it's, a marketing term okay, that's what it's become right there's a lot of. Buzz. Around it. Being. A practitioner, of it for 20 years I would, like to talk about in a very practical term what is it and what isn't it's in in the real world yeah. In. Practical terms that are breaking up in three concise. Categories, one. Is. What. We often think about which is analytics. Leveraging. Machine learning algorithms, and probabilistic, models. To. Look at unstructured. Data okay, things, like your customer data what are they saying what are they doing and, trying to get insights, on that data, that's, your analytics part let's. Use what people think about it's. Almost evolution, of your big data that we should be so excited about a few years ago and nobody speaks about anymore that's. What it is now it's it's AI for analytics okay.
The. Next layer of it I would talk about there's, conversationally. Which i think is a fascinating. Now. You're talking about leveraging AI to take the role of a human being in, an interactive, dialogue, and helping, a positon, solve a problem, very. Different than a chat pods very. Different than analytics, this, is about natural, language okay, and we've seen the examples of it but, we're not talking about chat BOTS that's something very different we're gonna talk a bit more about if you're interested and, then on the third layer I would break it down is, autonomic. Or, AI, for the enterprise, it's. About automating, the common tasks. In the enterprise, analytics. Data, conversationally. And looks at human interaction and then. Autonomics, or enterprising, looks at hey how can it automate the common, human, tasks in, enterprise that interact with systems okay and these can be things like interacting. With your managing your network or filing. An invoice maybe. Playing machine learning algorithms on top of your. Our PA and robotic. Cost automation, solutions things like this okay so AI, itself, I mean you've broken it down into three categories you've, got data you've got the. Actual human interaction and, you have the system itself, so, we'll, unpack that but I've. I've. Been in doing this for 30 years right so I have, great you just can't see it and. What. I've seen is every company I've worked at or every, company I've worked with has the same problems, whether, it's internal employee facing or, whether it's external customer, facing so. How. How. Do you apply this practically. And, maybe we should just go this one of the time right yeah tell, me how you would say okay I'm gonna put AI in the, CIO has just said you know what let's do this for data analytics, what. The hell does that mean so so and this is exactly the challenge most of, my. Colleagues, in the industry have is their. Leadership roles their, CIO. CTO. And, the. Board or the CEO right, says you know we. Leverage AI let's. Go we need to be ahead, of the competition we can't fall behind and, now. You're thinking what. Does that mean right right. So. Now you're being asked by, your leadership needs apply yeah your CIO has been told by his board, by CEO, yeah, go, play AI to my enterprise because everybody, else is doing it, well. How do you start the first key thing identify. The problem to ourself okay, matters board anything else if, you have a clear problem definition, you're gonna be able to apply exactly, what it is that needs to solve it that's right people, get caught up by the hybrid I to bring AI yeah, what you really need to do solve a problem you know let's let's go to that practicalities, yes, so I mean we talked about call centers and you know I look, at this as inside. Of a company, you know employees have, this experience and and I've had this all the time you know whether I'm going to I have to log in to workday to go look at something I have to go to Oracle, I have to go into a double us to look at instance sizes to you know there's. So many different enterprise systems, that are out there and you. Talked about a, chatbot. And in, this conversation or a call center type thing if I have to call HR after call IT or I have to send, in a ticket. Request the, experience is horrible in the efficiency is really really low so. This, is this is where I mean I think there's a there's a practical, implementation of, that digital, labor specifically. In that area and that's what you guys are focused on right you've know but you can go no why, why, is that different you know you're saying conversationally. That. Means that they can have a cognitive conversation. With you you, you've you've, hit the nail mat okay explain for, me you hit the most relevant question chat. Pods yeah, are. Very. Popular they're. Everywhere, there's, a tremendous.
Level Of, engaging. The market between them. They. Are not popular with your customers I do. Not like them I would. Ask your viewers yeah. Take a survey what percentage of them have had a successful, positive. Interaction, when you know it's actually a chat bot in a machine behind it that's, worse, I guess your type is upping they can't even get a context of what you're saying no, it's love chat possible okay. And. This is a point of our ticketing what is conversation, III and, in. Compare a chat bot because it looks the same right you're having a conversation with a machine yeah so what's the difference right. The. Challenge, that you have with a human, being is. They. Introduce, variants. By. Nature okay, if, I wanted. To do something. A thousand, times and make sure it was not done the same way I would, give it to a human being. Chat. Bots are rule-based, they're. Well defined, criteria, they. Look at buckets, and you fit in that bucket you do that one task yeah I, know, you must be equal C and, now if I, reduce variance. I have. Ensured, that I'm not gonna consistently. To. Define buckets and, it is very difficult, to find every possible variation a human being I think, it's impossible maybe, it's not but I think it's very defiant to every, possible variation, in. The way it will convey you want to make a transfer, from your checking account to your savings. Tomorrow, if you're bouncing over 500 ollars yeah say, that which other, you. Said so he will be yeah. I understand. The inference, understand the implications, are, based, on all that I need to check your balance I need, to check a time and if all these occur conditions, are true then IMing transfer. Contextual. Conversational, and eyes focus, is. Not one-shot interaction, yeah but, instead being actually, being aware of context. As you say to. Be able to have a multi. Turn dialogue, a multi. Turn dialogue is a, conversation, yeah like this it's exact right being able to recognize. The, intent. Ask. Questions, if there's ambiguity as. You do yeah get, clarification mm-hmm. So. You ask the questions you get clarification, you. Now understand. Okay. I understand. This person trying to take. An action and solve a problem right so you guys have you have right now the, ability to implement cognitive. Digital. Labor that. Can have these natural, language processing, conversations. With. Internal. Focused, employees, or external. Focus customers, you know and not. Right now we've been doing this for the last five years okay so this is not, we're. Just getting started man. Okay. We, are in version 3.8, of the technology, that. Is deployed at, enterprises. Across the world and this is this is both the chat. Right, interface, not a chat bot but the chat interface to be able to do that as well, as voice that's, correct it's on the phone yeah, it's on your home. Devices like your echo, or you know go home it's. On your mobile app and. It's on things. Like Skype, for business if you're talking to an attorney or. A. Web, chat UI get, them typically. So. Today. For. Example one of these, AIS, alone is handling. Hundreds of thousands of phone calls in, South America, from. A telco customer, it's. About I think I think yesterday had like 240 thousand, phone calls a day a day and there's, 240 thousand, calls it had about 92 percent completion. Rate meaning it's solved 90 percent of the call without. Vision I just I really want people to understand this the the. Transformational. Impact, of that statement that. These call centers with, humans that are unpredictable that are reading scripts that cause a really frustrating. Experience. For actual. Customers, suddenly, you've flipped that back around you can take even more and the efficiency. Of it and the effectiveness, of it and the customer. Satisfaction. Why would somebody want to talk to a a. I instead. Of a human the. Simple, answer has been this we've. Actually seen a 14%. Increase in, customer, satisfaction, so. AI wasn't, on par with human, support it was better yeah and, the reason was now you don't packet whine the. Reason was one you. Were never on hold yeah, you, called your. Answer was not. Five, minutes later -. You got a consistent. Response and AI, took action to solve your problem documented, it for you and if you needed human help it got you know and it can help the scattered, conversation. That moves to different areas right that's me so now that we're talking about conversational, AI that yeah any. Conversation, for you and I to be successful, I will candidly tell you despite. The smiley face I am a difficult, person, when. I ask for help I could be demanding, and I may have multiple things on my mind, I either help you with and in my mind they all make sense yeah but, I'm switching contacts yeah the, role of conversational.
Area Is to understand, the context, of the, conversation is having, yeah and then be able to say oh you. Started called me about making a transfer, and, then you guys you lost your credit card today. Obviously I'm sorry but how much do you want transfer yeah. Let's. Go solve your credit card problem right and after I solve you lost credit card and by, the way I double-check do you have disputes yeah are there any college of transactions, come back oh by the way did you welcome to your transfer that's great I understand your context, yeah and be able to apply it so today those are types of roles one, is as you said yeah, customer. Facing the. Other one we see as you rightly point out is employee, concierge, and but. Can you consolidate think, of a large enterprise you. And I have both every. Authorized. And. How many different contact, points are there, employee to try to get something but it's so frustrating and, again from personal experience, it's just is that if. I could make it really simple like for example take, what you said if I had one chat interface to be able to say I want to do this in. My own natural language, of the way I type and it, can unpack it and do it oh so. Much easier because they don't file a ticket I don't have to follow up with somebody to call whatever and it is that done it's, actually happening that's a huge transformation and, you skip the first step figuring. Out who the right person is great to get with true, which is the, first block of anything yeah if you've noticed that most enterprises, the executives, will. Go call a person, they know who, knows. Make. This happen for. Every. Employee will have a. Concierge. And, this is where you think about a. Multi-role. Conversation. Later yeah so you now think about and we have deployed these to put a very large Bank here. Number. Of our large banks in the US a, large. Insurance. Company, what. We've done is we said okay, let's. Have each. Of the different lines of business in the firm train. And. Educate the conversation, yeah in their role so you have an HR yep, conversation. Is yeah HR do you have an HR finance. You. Have one for finance you have one for compliance. You, have one for IT Service Desk each. Of these Emilia's have. Their own hat their own skills yeah but, in one conversation, with you she deserves based on the context, which. Part of her knowledge that she need to use so you never have to know oh that. Question, was actually answered. By compliance, yeah not by a chart right that could be a blurry line yeah, but, the machine is aware, of different leverages. And asked the question you can go across those domains as you go that's amazing so how come BB King Saul Paul I've, been in okay so let's shift gears here because, I think what we've explained. Really clearly is that there, is an opportunity both, for internal facing. Interactions. With humans right, whether employees etc, and systems and there's an external as well but, now let's flip it around the people that have to do this the system, solution, itself, that next category, what. Does that look like on the back end how does that work you you ask the right question and by the way it is a challenge. Most, enterprise has been actively, trying. To address for a very long time. Every, enterprise that you know has either. Bought software, or built software or a combination of the two yep to. Help them manage their, infrastructure. Their technology. Their data centers. Their. Applications, etc. You. Have these ticketing solutions like ServiceNow, your monitoring, solutions, you have accept, we all happen yes I can't imagine one enterprise today unfortunately, I can't. Imagine that one that prices that doesn't have more than one of these solutions right, now yeah that's, a separate problem I will try not to judge, but the reality is you, have all these piece of technology mm-hmm, okay, great. You. Them still. Have all. These, people, that. You've had to staff, to. Fill the gaps, and. Middleware the empty space between, all those pieces of technology, you brought in you. Have a ticketing system your monitoring system and yet people in between doing. Those things are trying to act on that yeah, and they maybe have our paper here that there's one part of the process, to automate like all the way on the edge but. And you said rpm yes the robotic, cost automation, so so yeah and it's, basically. Automation. It's basically, specifically. So I think of automation in three layers if I was to break it down, one.
Is You can automate through ap, ops so, programmatic. Interfaces somebody's, written a system, you can call directly yeah -, you can automate, through the command line like active. Engineer, and. Like. A system in your database engineer and log in and do something directly on the server sure and the third one is you need to do it on a. Graphical. Interface like, the desktop interface a graphical user interface to, do that that's. Our PA so. A first layer so the first our API is, to be the integration framework doesn't require a PA, command. Line automation, that's. Where we use a deterministic. Finite state machines. To. Be, automating. The command line yeah and, the third one our PA is when I know you need to open up a desktop a browser click, around and do, something yeah so, those are the through there so now, you. Typically, at Enterprise you have an RPA solution, that's, doing something on the screen, but. The small part of a problem, it's while you're doing that narrow, definition, of the problem yeah you, have some scripts. That you've written here like, PowerShell, scripts to automate on the command line yeah and, then. You have all these API, what. You really need is end to end yeah automation. And system. Automation is about saying I don't. Care which. Of those three I need to leverage I don't care which combination. Of those three I need to leverage my. Job to solve the problem yeah so when I get a contact, from a customer, or when, a customer's modern system sends me alert saying hey I have this problem the. Enterprise, autonomics. AR says. How do I use the tools available to me to. Solve the problem that came in yeah and now it will go reach out to API as needed it, will run, an autonomic. Engineer, to go do something on the database yeah and if needed, will go run that RP I pop our, mandate. It is this is a the. Problem we focus on with one desk. Is. Can. I layer. Enterprise. AI like one desk across. All the existing tools I have and leverage. Them so, the AI provides. A value, in automating. Operations. Without. Having to replace the same piece of software yeah yeah widely across this business is so important at this point you're saying because I you. Know the the I've. Done this a number of times it's very complicated, especially if you're trying to integrate companies, or you're trying to split companies, or you, know just implement, those systems, it's, all those disparate systems and, so having the ability to aggregate them. All together into one way in which they're interacted, with right. That's the punchline of this and then so let's let's talk a little bit more about the, you know how, how. IP soft does this and, not the technical underpinnings I'm saying that we've got those concepts, the what. Are the solutions and. So like one desk and one store etc what, would people use, to. Do exactly we've been talked about so. Let's take from the top down. High. Level. One. Desk, contains. Amelia which, is our conversation okay, one desk is the enterprise AI okay. Its job is to drive. End-to-end. Automation, in. The enterprise. Leveraging. All of the spread technologies, they have got, it and just to be clear so if one desk is the interface, Emilia. Is the AI on the background so, it's Amelia it's a part of the air okay so the to try, and define. Its well. Amelia. Is it conversational yeah yeah Amelia's, job is focused on being able to understand, a human being right solve, their problem in a voice or chat nailed it yeah in, addition to Amelia you have autonomics, AR I got it which is about being able to interact like a engineer.
On A command, line and solve a common in. A sense of interaction, issue and then, there's our PA AI which, is one our PA yep, whose job is to be able to leverage AI and machine learning on top, of desktop. Interactions, except, taking, all this contains. All three of these a, combination, of these and when, you look at where. We, would focus your mandate, if. You are focused on. Solving. A call center. Problem. Where are you trying to deliver more efficient call some experience, expand. The coverage provide. AI, in the new channels that's. Amelia mandate, and. I have very strong advice on how you would identify use cases there that are specific, to well make successful error if. You're looking at I have. An enterprise, organization. And I have, HR. Finance, I, have IT and I, want to run my enterprise, operations. More. Efficiently, yeah that's where one desk comes in got it one desk job so then help you look at your enterprise, as a whole leveraged. Emilia, autonomics. And AI base or a PA to help, you automate end to end inside, the enterprise Emilia. Is conversational, it. Can be internal external one. This is about automating, the enterprise operation, itself across, towers yeah, okay so, there's practical, solutions, and you have this new thing called one store yes, this. This is what blew my mind when I came back and started talking to you guys again right, I just look digging in deeper the fact that I can download digital, employee. You might get the AppStore well. You've. Said it very well you said it very well, that, is, probably, the, biggest value, we've seen we were able to bring our customers and. I'm a geek that says a lot because I really like a technology, yeah, one story is not so much about technology well, it was about experience. In. The market today how many companies are there that, have some where AI in their name. Okay. What's, the average age, well. Those of the companies yeah oh yeah they're they're not that old a couple, years yeah maybe laughter, yeah.
We're. 20 years old. For. Good or for worse we've been applying AI. Solutions. For. Our customers from, the foundation, up yeah it's a decade yeah that's that's fascinating, and. As a consequence. Of that and I'll be very candid it wasn't planned but an, unplanned consequence, of that is now we have, when. We began I would, come to you in mr. Nelson I'd say I have an AI platform, it will help you all drive, automation, in your enterprise across, these different powers, and their ability. Today. I now. Have 20 years of trained. AI. Agents. Yeah, the. Foundation. Of one store and. It's. Really we, think about as a digital workforce. Marketplace. Where. You, can, go and hire. Digital. Employees, that have years, of experience having. Done that job at other enterprises, bring. Them in so you can bring an HR, yeah. Digital. Employee or you can bring a Oracle, expert, digital, employee for your database or a banking, expert for your customer facing right, so tomorrow you're not starting where, I was at the beginning when I'm giving you a solution and say let's figure it out together yeah I'm now, saying, here's. Your Baker right, start, with your banker that understands, in terms of terminology if, you say balance it's not about standing on two feet it's. The amount of funds. In a particular account yeah I understand, the language under steady actors no order transfer, means and now. Start your customers off and that's what our focus has been now is with one sort taking, that 20 years of experience that we have, bringing. Them together and preach train enabled. Digital. Agents across the technology across conversationally. And I across. Autonomics, across our PA and. Then being able to make them available to our customers in that online, format, so they can bring them into the escrow, and I'm gonna get down to two last questions yes sir um now that we've got all these an. Understanding. Of the systems and how this is you know the underlying the underpinnings, and how it would work can. You tell me some of the results that have happened with people who have truly implemented, this yes at scale and, also if you don't mind my lessons learned yeah because I have made mistakes, no, that's shocking, I know 20. Years what could possibly, go wrong I figured, that out every possible, way. When. You're looking at. Applying. A nine my first advice was is the same as my last advice solve. A problem yeah, that, also means don't, use a young i4, FAQ, don't, use AI for civil use, ionic. Solutions, to actually take action, solve. Our problems I have, had my poor side experience where customer want to move very quickly. Then. Integrate. With their back-end systems, we said fine we'll, do an FAQ based, conversationally. I achieved. I believe, 92%. Accuracy, with. About 24, percent customer satisfaction okay. The. Other resounding, success right. We've. Not taken the same solution. Enabled. It with back-end systems. To. Actually take meaningful action. You. Don't want, 240,000. Calls are answered in the day by a AI because. You can actually solve the problem yeah, focus. On the problem, that you're trying to solve and enabling AI to actually, take action on it when. We have seen that with a focus, we've. Seen customers get. Range. We. Had all states they, published Wall Street Journal article on us yeah a blog. Actually I believe.
Within. The first, deployments. Their, first time right one from 66, to 76%. Okay. That's, a goal life think. About over time mm-hmm. The average call handle, time went, down by, more than 15% I go, alive and. Trending balances. These are the type of metrics that you see but. It, requires, you to actually, focus, on use keys where AI isn't able to take meaningful action, because. If you just deploy your AI. Without. Enabling with the IT systems, are back it's. Not going to drive in value yeah but if you do then. You can actually see solve real world problems and relatively, significant. Metrics awesome. So hundreds. Of thousands of actual, voice calls a day absolutely and, we're handling, I'm. On the digital channels I mean we were somewhere, over around, in. Months. In a month paper millions, of chats yeah, Hannah biak wants multiple roles, we. Probably have, today. Driven, by customers, some 400 skills, different. AI abilities, are enabled in the market and continuing to expand. We've. Seen one. Of my large client service firms reduce call center costs by about 72%. One. Of the five things you call your call your tell about your wireless provider about put, five my, bill reception. I'm traveling internationally changed. My plan yeah if you handle the your your, Pareto principle another, that very wise Italian to gave us a to a rule if. You apply to 80/20 rule, focus. On 20% of your volume that dress 80% of your. You'll. Actually address quite a bit usually, these these. Opposing. Things, happen you try to get more efficient, and you have a worst customer experience, you, try to have a better customer experience you drive down, efficiency, and increased costs this. Is the ability to actually solve both together as long, as you enable, the AI to solve, a problem yeah, give it the fair playing field to actually be helpful that yes got it so we're. Gonna wrap up with one last question yes sir if. Anybody. Wanted to take one, thing away from this entire conversation what, should it be. Focus, on the outcomes okay. There's, been a lot of hype around AI. Which. Means every Enterprise wants to try it bable their toes, on it but. It's very easy to get the solution with it mmm-hmm that means if you try to bring AI and, we're, play with it it's, very easy to lose focus from, the rest of the organization, because they say you know what it's science fiction still and there's in a real world yeah, I would, focus well defined problems things, I understand. What that, I have measurable outcomes against and, will apply my a solution, to those measure palms because success, is your. Primary catalyst. For driving adoption especially. In the market of AI where there's so many questions about it and real it'll understand so I would focus on excellent. Outcomes well Aragon thank you very much for spending the time with me I appreciate it the pleasure awesome. And thank you very much for watching and we'll see you on the next episode. You.