Sign Iowa Banking Living Will Simple

Industry sign banking iowa living will simple

okay welcome everyone and thank you for joining myself siren dial i'm the ceo of majia solutions and barry hoon who's the director of product management for the oracle digital assistant based here in australia we're going to talk to you today about how chatbots can be used for industry and one of the industries that we're looking at is the utilities industry but equally we have implemented chatbots across a range of other industries including higher education government and retail we'll show you some examples across those industries today so welcome and thank you barry for joining me it's the obligatory zoom are you on mute thank you for having me no worries so it's good to see you again barry okay just a little bit uh about majira to begin with i won't spend too long so you know who we are uh we are an oracle specialist partner we specialize in cloud solutions from oracle we've been a platinum partner since 2017 and we're also recognized as a certified excellence implementer which we're quite proud of because that is judged not by oracle but by our customers and oracle's customers and i think we're the only one in australia who has that we're fully australian owned and operated with staff in all major australian cities we do do work in the region as well outside australia so new zealand and broader into asia-pacific and in fact one of the customers we're going to talk about today is an offshore customer for us we made the pivot to the cloud uh really back in 2012 because it was it was pretty clear that was the future and with eight years of experience under our belt that's that's just picking up momentum all the time and in the last fiscal year we grew 60 year-on-year largely on the back of oracle's growth in software as a service we can deliver any oracle sas application you can go to our website to see them in detail so that includes the erp suite so erp hcm and scm and also the cx suite and today we'll be looking at one of the platforms of service offerings uh the digital assistants but it's very common for those to be integrated with the other sas products in particular the cx suite and oracle itself has done a lot of integration there so one of one of the big areas for that is the integration between digital assistants and oracle intelligent advisor for example which is a natural language rules engine that lends itself very well to a conversational interface from a professional services perspective we offer the full range from implementation through to managed services and we can equally do technology and business strategy for any prospective customer okay so getting on to our topic today when is conversational good in our view you need to have some important information that a customer can access through the conversational interface things like what's my lead balance how do i do something what am i eligible for similarly transactions are good help me apply for something update something for me take me to checkout and finally interaction particularly with the devices so being able to upload a photo or submit a document place an order these are all things that that lend themselves to a conversational interface so the main caveat is it's only worthwhile if it does something useful and therefore the other key thing we see is that ease of integration to source systems is very critical and oracle has certainly done a lot of work there to enable that integration to make it smooth and make it relatively easy for a partner like us to implement for a customer we think there are two key elements to bringing these things to life the first one is figuring out which channel you want so do you want facebook slack mobile applications wechat any of the conversational interfaces that particular gen z uses and i think you'll hear from barry a bit about some of the drivers for uh opening up this product having you from oracle and then the second thing is well what's the intent what do you want to enable with the conversational interface a chatbot with no purpose is kind of like a robot without a home so again that comes back to having useful information that can be surfaced through the chatbot coming from host systems of record okay barry i'll hand it over to you here to tell us a bit about all of the work that oracle has been doing to make these digital assistants worthwhile thank you for that let me just pull up the slide and make sure i'm connected to the right thing okay so thanks that surrender um yeah so what we're going to be introducing today is oracle's digital assistant but first of all is that i've been in the sort of i.t industry now for sort of since the dinosaurs and what we've seen obviously as people moved from the traditional client server is every 10 years or so we've had a very significant shift in how users interact with a company or interact with their sort of personal information and obviously back at the the sort of turn of the millennium we had the web and everything was about building web pages everything was about building web pages and we really did get to the point where we almost had too many web pages to deal with but it was very hard to find the answer and then along in 2008 or whatever it was the the iphone came along and suddenly it was all about apps and everyone needed an app and it's like i need a mobile app for my customer i need a mobile app to do this that and the other and it's like well what do you want it for oh i don't know but i need a mobile app and we're now moved to the next generation which is really this idea of the conversational interface and as we'll see one of the downfalls in mobile apps was the sheer number of apps but from a conversational interface people were rushing to build one of these in the same way as they rushed to build mobile apps and in many ways ran into similar similar issues but when we look at it why do people look to conversational interfaces well there's a number of reasons why people see these as particularly useful for customer support and employee support and and first of all there's no app to download it's not something where i have to proactively go to the app store and download the app for something that i hopefully wouldn't necessarily need so in the case of needing to report an incident i probably wouldn't download the app just so that one year day a year i would actually need it so the advantage of chat bots is they're using apps that people already have or they're using interfaces that are already available be it things like facebook messenger or wechat or slack or whatever depending on whether it's sort of inbound or outbound the interface is already there you're just taking advantage of it so in other words you're exposing the organization to many more channels by effectively rather than have the user come to you you've gone to them and it's using a technique that everyone has been doing ever since they were small children which is basically being able to have a conversation so rather than having to learn an application or learn a mobile app effectively the advantage of a conversational interface is it's really there to understand what the user's trying to achieve and then hence respond accordingly the downside of course is with conversational interfaces there is lots of different ways you can ask a question and this is particularly true if we're dealing with some of the earlier attempts the the series the google assistants of the world where it was really set up to answer a specific question and if it didn't understand that question it pointed you to a website with many many links to say well i can't answer that now but here's hundreds of links that i found that might or might not be any use to you so one of the downsides of that first generation was really the experience where it was not about the conversation it was about answering a question which didn't give you to serene's point it has to do something useful and if it can't take that next step it's not useful and one of the things that really makes it difficult is the fact that as humans as we speak to each other we don't actually speak in a linear fashion so even if i had a process i don't necessarily go a b c i might actually go a b c or another thought occurred to me i'd ask a different question and then i'd go back to the original question with the assumption that the person i'm talking to hasn't completely forgotten the question i have posed in the first place so this non-linear is one of the things that really makes or breaks a lot of these conversational interfaces because people won't stick to what's effectively referred to as the happy path so let's start by sort of differentiating we talk about digital assistants and we talk about chat bots but really what is the difference what what makes a digital assistant if you like the next generation of chatbots well as i said if we looked at the likes of siri they were designed to answer a specific question and have natural london understanding of that question chat bots were about a use case so for example i would start a conversation and i would build a chat bot around a particular use case or a particular task so this might be my service desk chat bot or this would be my billing inquiry chat bot or my employee leave application chat bot so effectively you ended up with a chat bot that was really designed around a single use case and the reason behind that was very simple the state of the artificial intelligence was such that if we try and put too much into the one bot it gets confused so it's not able to work out do am i talking about topic a or am i talking about topic b so what a digital assistant really is is that second generation of chatbot where the digital assistant effectively acts as a master bot or a super bot that sits in front of these individual use case bots that we refer to as skills so effectively each one of those hexagons represents a stand-alone chat bot that understands its own business domain that understands the the conversation required to achieve a specific goal the digital assistant on the other hand sits in front of that and is able to route to the appropriate use case based on the incoming question so this is where i said we'll introduce the oracle digital assistant and said the digital assistant is one of these master bots that allows us to have different skills that are agnostic to the back end application whether it's an out of box application a bespoke application a third-party application effectively build a for whatever better term a smart lego block that self contains all the information about that need for the use case and we simply plug them together to create one of these master bots that is accessed from a single point of contact so effectively even though we may have multiple use cases from a user's perspective it's a single experience regardless of the type of question they ask if i'm asking a business question it'll fire the business use case if i'm asking a a leave balance question obviously as an employee we're able to bring that back in the same same conversation flow but as i said the thing that really will differentiate a digital assistant is also this ability to if you like go off topic to change the process you're doing so as we were talking about before the bot needs to do something but if i'm going through a process so for example here we have a new hire who needs to order a new laptop so they're ordering the laptop they're halfway through the procurement process and it's like oh when do i actually start again so in this case we should be able to jump over to that other question answer it and then come back to our original task and complete it accordingly so in other words that non-linear approach or a non-sequitur that allows us to effectively jump between thoughts that the user's having in a more basically humanistic way but when we design a bot there are actually different types of interface that we might want to implement and as we'll see in a minute there are different things we might want to plug in to achieve that so a transactional chatbot where we're going off and getting things like account details or when's my build you and things like that are effectively faqs but they're faqs about my information so they're very personalized and we need to be able to go into the backend system of record and get that conversely in a classic faq the answers are pretty much static based on what the question is being asked by the end user so in that case when we talk about faqs we're really looking at global information where based on the question we can go and get the appropriate policy or whatever from our knowledge base likewise there are many cases where you need to have specific uh guided conversation in order to pass a business rule so if i'm applying for a new phone or applying for a new account there's going to be very specific questions that as a user i need to be asked and hence this is where the concept of guided conversation which where intelligent advisor and we'll talk about that briefly is able to effectively become part of the conversation so we can look at a conversation being user driven or application driven so in this case the user may ask a question once we've worked out what they're asking for the application may take over and subsequently guide that conversation to achieve the ultimate goal but realistically you're never ever going to be able to build a bot that does everything and one of the things that again that can make or break a successful customer service chatbot is the do you have the ability to effectively hand over to one of your call center agents if required and we'll look at that in a bit more detail but if we look at today's customers needs there's really broken down into three areas instant access to their information rather than leave them on hold for 45 minutes or so with their call is important to us let's keep them happy by giving their information straight away likewise make sure they have access to the appropriate information for the question they're asking and for who they are are they a customer are they a vip customer are they an employee let's make sure the answer is right and then i said in terms of building their accounts building things having that personalized guided approach to actually make sure that they're quickly and easily achieving the goal that they're trying to achieve by coming to your to your site in the first place so obviously in our case that instant service is really where digital assistant comes in but if we're also wanting to have an faq we can take advantage of various knowledge bases to have that sort of standard policy questions available to us so we can surface those and as red mentioned intelligent advisor allows us to do guided conversation so if we look at it the various components make up this digital sort of um triumphant if you like where we have base knowledge whether that's sort of uh policies or standard faqs you have the conversation flow and you also have guided conversation and the three components fit together to build up this sort of overall sort of conversation flow that allows your customer to access the right information that they're after and do it quickly so how do we build these things well especially when we're looking at transactional applications digital assistant and oracle has taken a very different approach to many of our our brethren in the industry by looking at how we use artificial intelligence so rather than assume that most companies have data scientists available to them to build up the the ai or understand things like tensorflow and the various models you need to to actually build an application oracle's taken the approach that we will build the appropriate ai functionality into the use case so in the case of digital assistant the artificial intelligence is all about natural language processing and conversation design and it's built into the engine so that way when we come to design a conversation it really becomes a business function it's not an it function because the business understands the types of questions that a user is going to ask they understand the types of of sort of utterance that is going to go with the domain and hence they're able to build that up as a sort of almost like a meta design whereas when we get ccess from the back end we still need it in there to say how do i get to my customer management system how do i get to my billing system etc so i t in this model it handles the integration side but really the business and the knowledge designers build out the conversation flow because they're closer to what the customer is going to expect and the types of utterances or the types of examples that are going to be needed so the way we build it is it is a very declarative model so from a user perspective they'll go in there and effectively what are the user goals that the user came here in the first place so in this example here which is actually a banking example they may be looking for an account balance or i need to do a transfer and then we represent how that user might ask that question so how much money i have is indicative of i want an account balance once we can have that we can also tie that to important bits of information that we might need to capture so my account number my account type am i uh a sort of in the case of until a phone network am i a broadband am i at a landline am i a mobile so basically the type of information you can define is again defined declaratively with examples of how people might ask for that and then having got that the business can simply sit there and ask questions and see whether they've got the right information there and that way they can retrain it or or change the behavior of the bot in real time by saying if i ask this question what's the outcome but the other side of the coin is the conversation itself so in this case what we have is the ability to define a conversation flow based on if you like almost conversation by example so as a business user you would come in and say well what did the end user ask what would the bot respond what would the user then respond to the bot and basically build up that conversation he said she said and use that as a guiding factor for how the conversation would be and in this case obviously what we're really looking at is the happy path which is the successful goal and the engine itself will then take into account if the user goes off topic but as i said there's always going to be a situation no matter how good or bad our bot is that we're not going to be able to answer everything and that could be simply because we've got the wrong um we haven't got the right domain in our bot it could be that we the user asked a question that wasn't um in the system or it could also be a purely pure personal thing it could be that the end user is upset the end user didn't understand the bot or simply that they just don't like bots so they want to talk to a real person so one of the things that can make a customer support uh chatbot more effective is that ability to hand over to a live agent as part of that conversation flow and make sure that the agent understands the conversation that the user has had with the bot so in the case of oracle digital assistant we actually allow this in two different ways and the use case really comes down to the nature of your help desk environment so one of the areas is allows us to place the bot effectively in front of your help desk so in this case the bot starts the conversation with the end user and only if it detects that there is a need for an agent whether that's implicit or explicit does it effectively escalate to your call center so in this case obviously we're looking at say oracle service cloud but it could also be other call center environments so in this case you're escalating and you're effectively handing control over to a live agent that takes over the conversation at that point so in the first case it's human talking to bot and then it becomes human talking to human simply by transferring or escalation from the bot itself so within this as i said it has the again this is a very declarative approach we allow you to say where is my help desk uh who do i connect to if i hand over and then that way the bot is able to hand over to the appropriate back end and as i said the whereas out of box it's available for oracle service cloud we also support third-party systems like zendesk salesforce genesis etc where we have the framework that allows you to plug that in if you so choose the other side of the coin is really to take advantage of the functionality of your help desk itself which if i've built specific routing rules into my help desk where the user comes in and they're asking about a particular issue because you've gone through a triage then you may want to route to a specific uh either the bot so initial questions go to the bot important users say this is a vip customer or something like that they may get routed to a human regardless but in this case assuming say the the majority go to the bot effectively the digital assistant is acting as an agent inside the help desk environment so you're still using unlike the previous version where or the previous option where you were using any channel available to the bot so if you wanted to expose your help desk through whatsapp or or wechat or or facebook or whatever you could do that in this case you're using the standard help desk client and the bot is actually sitting behind it and is able to transfer as required so i said it has the advantage of using all the knowledge that you may have built into your help desk environment and the digital assistant or the bot is really taking on the sort of the ground noise question the repetitive questions that are easily answered rather than waste the time of an agent you can help have the bot take them on uh and process them and only transfer to an agent if it's absolutely necessary once you've built your bot how do you know you've actually built the right thing and this is one an area where analytics becomes extremely important being able to see the types of questions that your users are asking and have you actually built the right bot so if i build a perfect bot that is answering questions a but we're finding that 80 percent of our incoming requests are about queue then what that means is we need to modify or change our chatbot very quickly to address that so getting real-time view of the types of questions people are asking where are they coming from are they coming from the website are they coming from say facebook are they coming from for example wechat so in this case we can see where they're coming from and if we see the majority of people are using a particular channel we can also optimize the bot to be more effective on that channel but as i said on from the bottom left-hand corner here we can also see what were the top goals that someone came to the the bot for so what were their top intents or their top goals how many of them actually succeeded and how many of them actually sort of left or abandoned again it gives us the option to optimize that conversation to make sure it's useful for our end users and that extends to how often did we really need to escalate to an agent so if i'm looking at my call center obviously deflection has an immediate impact on return on investment obviously call centers cost money both from an agent salary viewpoint but also telephony and all the other costs so if i can deflect even 30 40 percent then i have a significant bottom line hit so being able to see real time how many were deflected but also how many then needed to go back to an agent anyway we can actually work out the true value that the conversation is bringing to the picture but moving forward we've talked about the conversation but the next generation if you like of interface is also ultimately going to be no interface and this is where obviously voice interaction comes in and what we're seeing obviously everyone has a mobile phone that has siri or google assistant or they may have google home or alexa or things like that but the the function or the problem with those environments from a business perspective perspective is where that processing takes place and who owns the data so oracle has introduced voice capability that is very much designed around two main factors security and also understanding your business so obviously if i'm using a google home or something like that every time my user asks about their account number it's going to a third party and that third party has control of or logs that conversation in the case of oracle the voice capability here is within your environment it's within your instance of the bot so it never leaves this environment and in fact doesn't even record the actual outcome or record the message so this means that you have much greater control over things like personal identifiable information because it stays within your security domain but the other side of the coin really has to do with terminology it has to do with domain so when we talk about automatic speech recognition there's the concept of open domain hey what's the weather like so that's something that is realistically a general conversation but if i was to ask what's the who's the key account director for one of my key accounts key account director might be fine but here for example in oracle key account directors are just colloquially referred to as cads so who is the cad for our particular sort of utility company that we're talking to say for example guam so in that case cad if i was to ask one of these open domain people who a cad is it's just as likely to come back with canadian dollars so this is why it's important that we have an environment where you can effectively teach it new words that make sense in your business domain and speaking of customers we have a significant growing base and as you can see there's a number of large utility companies in there be it both electrical um as well as things like uh gas and oil and things like that so in terms of growth using it both for customer service but obviously large corporations have large numbers of employees so being able to give your employees access to the information that they need to successfully be an employee makes it easier for them too so it's not just outward bound inward bound can have a significant impact on how employees are able to access their own information be it hr information be it job function information so just to give you some ideas here so for example a good example would be comed which is part of exelon and they're using the chatbot as a way of servicing their customers to give them access to things like billing information being able to do reportage of outages etc and and looking at the broader company which is exelon we can see that they've had a significant impact through deflection um and been able to support more customers than they were directly through the call center so effectively they're getting a 30 reduction in the cost there simply by being able to have those customers go to the chat bot rather than sit on the phone waiting for the customers to ask about their bills asking about the the fact that they don't have electricity or whatever likewise we can see with in japan the tokyo gas company had a similar thing they went through a social media approach using line which is one of the dominant social media platforms in in north asia and looking at rather than being tied to when the the the call center was open really looking at going for the 24 7 and again by having that deflection level they were able to make significant reductions in cost simply by servicing those customers immediately rather than having to go through a call center and obviously we have others like telephony phone companies and in fact one of the the biggest chat bots uh around um phone lines and is actually based here in asia pacific our the philippines long distance telephone company um it uses exactly the this approach to deflect from their call center and have had a massive move to using the the chat bots as a way of allowing their customers to get access to it straight away okay so at that juncture i will pass you back to surrender all right thank you very much barry uh let me share my screen so i can talk um and thanks thanks very much for the both the product background but also some of those international examples so i'm just going to talk a little bit about local examples things we've built with oracle for local customers and then give you some views on how you can actually get started if you want to do one of these things so just going to pick three example use cases locally that we built the first one is dhl australia and we implemented for them a full electronic commerce solution and it's to allow the pharmacies in particular and the hospitals that they supply to uh to order and buy and uh and track everything online and that's a pretty important function in australia particularly at the moment with covert because these are mission critical suppliers if the hospital doesn't get the things it needs then people can die so the big thing that the bot does here that the digital assistant does here is provides tracking and tracing so it's like where's my package and customers can see where their package is up to and do all of that through a conversational interface or equally they can do it through the normal website but the conversational interfaces uh has been very popular for them and in fact i was talking to a pharmacist uh recently who just said look it's like night and day now we can do it all digitally we can see what's happening i can go on my mobile phone and just check through a chat interface where things are up to so they love it the end users love it of dhl i can't show that one because it is kind of secure access for the kind of b2b although it's almost better to see because the pharmacists are very small businesses the second one uh which i will show you shortly is from the university of tasmania and they've actually named their digital assistant mumford and it provides information on fees personal details results etc and you can upload photos and documents or share your location with the digital assistant and for them the big driver for that is uh student engagement so the students can very easily interact with the university through a simple channel of choice and particularly for that generation of gen zed chat is one of the main ways they interact with each other so it's a really nice way for them to interact with the university and then the third one is the guam power authority so for them it's transactional information how much was my bill when's it due and they can then get routed to a transaction and in this instance it's integrated and this is one of the real strengths for oracle particularly in utilities uh and i think a number of the people on this webcast may be from that industry who have oracle's utilities software so the system of record on the back end is an oracle system for um for their utilities bills and managing everything to do with that business so that's where the information that's where the digital assistant is getting the real information from to help an end customer so let me just bring up mumford if i can work my screen properly uh here we go so this is you can see here mumford is the um the chat board intern who's joined the team and these are the main topics or skills if you like in various terminology uh that the uh that mumford can help with so i can ask a question like i would like to enroll and comes back and says okay i've got some things here how do i enroll i'm having problems enrolling what do i enroll in uh let's say i want to know how do i enroll then i get asked what do i want to enroll in do i know what i want to enroll in if i don't then it can give me some guidance about that i'm looking for an elective or experience an engagement unit i'll say yes i am and then i get taken to a unit search and if i follow that link uh it takes me off to that part of the website uh so there's a a whole range of things as barry was saying here it's like multiple bots have all been aggregated into mumford for the students to interact with the other thing that i can do is i can share my location with the bot and so i can get context sensing information to where i am now i'm not in tasmania at the moment so that might make it a little bit difficult for uh for mumford to help me about that again okay yeah i'm kind of out of the boundary for uh for mumford usually it's helping students kind of near the campus uh the other the the final example the guam power authority here and i'll just briefly show you the screen i can't i can't do a lot with this because i'm not a customer of the grandparent authority but you can see here i think the nice thing here is it's integrated on their facebook page and you can ask questions like when's my bill due how much do i owe and then if you're a customer you could it basically just gets your account details and provides you that information from the host system of record and for these guys they've it's reduced the the call waiting times by about 45 uh because i think barry mentioned it earlier the uh the one of the big things with this is is deflecting calls and deflecting things from the call center so that a lot of those basic inquiries people can just do themselves now and then it's only the more complex ones that the humans in the in the contact center have to deal with uh and then for mumford the other the other thing uh to point out here is the the systems on the that sort of sit behind that they have oracle service cloud so you can do that switching between a human and a bot uh as needed for the uh for for whatever the end user is is looking to get okay so if you want to get started with these uh we at mcg we've got kind of three starter packs essentially three ways to get going and they're really not as expensive or difficult as you might think to get a digital assistant up and running the first one is a starter pack and that is basically a proof of concept level project where you're looking to test some basic applications of the technology you get one environment the key thing really is integration points how many backend systems do we need to integrate to you get three medium complexity intents uh and and a single channel so it could be like facebook or the web for example that starts about 20 000 and it's live in two to four weeks the next level up the professional so if you still want to get live quickly and then kind of maximize your time to value you get two environments here still one integration point five medium and complexity intent so a few more two different channels so you can go like facebook or alexa or the web uh and that's it that's about thirty thousand dollars and you're live in three to five weeks for that one uh and then the the largest one is that what we'd call the enterprise and that's where you've got more complex requirements and processes for this one you get two environments three integration points up to ten intense uh and multiple channels up to three channels and this is around about uh fifty thousand dollars live in four to six weeks so any one of those three is pretty fast you're going the biggest one is sort of four to six weeks the shortest one could be two to three weeks so thank you very much everyone for joining us today thanks for listening um if you want to get more information please uh reach out there's my linkedin details there you can visit us at our website or send an email to info magee solutions.com we could we can take questions if anyone wants to just ping them through the chat interface i'll give people a minute or two if they want to do that feel free to ask questions we don't bite yes we are friendly if you're interested say for example in the comed one you can actually go directly to their website and and have a look at their bot there but as i said one of the the big things with any chat bot is that deflection so if you can make or prevent your customers sitting on the help line and get their answer straight away you end up with happier customers and in many ways you end up with happier call center agents because they're focusing on the the the issues which really allow them to show their professionalism um answering how do i reset my password is probably if you've asked that question 80 times a day might become a bit repetitive so that's the sort of thing where the bot can take over yeah absolutely all right very good it looks like people aren't keen to to ask questions over zoom that's fine um feel free to uh to send uh to send questions through to us on on any of those uh on any of those channels there but we we think it's a you know very interesting offering from oracle and we're very well equipped to help people implement it oh here we go we've got one for you uh barry how does the chat through the complexity of the language that's a great question okay so um again that's one of the the big things between the sort of early chat bots and the latest generation of chatbots and let me just bring my video back up again so you can see me um so the the nlp or the natural language processing um for uh the conversation is basically a using neural net so it's a deep learning model so the nice thing here is it's able to when we train it we don't have to give it huge number of examples we can start with a very low number and the nlp will allow us to understand what the question is or do what's referred to as intent determination so once we know the goal of the user we can then look for the specific information that we need to find so rather than sort of look at every possible use of a word we determine what the user is after first and that automatically scopes any other information that we need to look for so in terms of the language it's not like the early generation nlp that was rule-based or using grammar this is a machine learning model where it's using your utterance or your question as an example and then it uses that deep learning to work out ultimately what the user's trying to ask okay if we want to integrate with wechat messenger is an additional cost um obviously from a viewpoint of so facebook messenger is out of the box uh wechat integration does require obviously a wechat business account so they're the cost is so we can do those certainly in the in those starter packs basically you get one with the starter two with the professional three with the enterprise and you can choose the channels for those there is a little bit of work as barry says for the ones that aren't out of the box the um the then we just have to do a bit of integration work to get them up and running okay um and in fact we have we currently have about a dozen um sort of channels directly available uh but then as uh siren was saying we can expand on that to to have other channels so another question here um what are the scenarios where a human agent will be needed um that really comes down to as i said three main areas the the first one is you may not have designed the bot to answer the question um so if they came in uh and well-designed bots we'll we'll take random questions and we can look at random questions in in two different ways or in fact three different ways is it a question that has absolutely nothing to do what we do as a business so as a utility company if people came in and said oh i'd like a pepperoni pizza pizzas are nice um we don't want the bot to understand pizza because it's not our business domain so in that case um we we just want to ignore those you've got the next scenario where the question is um part of your business domain but not something you've currently trained the bot in and the third one is it's actually part of your business domain and part of the bot but it's not an example that the bot knows about so in other words that becomes a training issue so when we look at when do we transfer to an agent we only want to transfer to the agent basically on three areas one the it's not something the bot is actually set up to answer two um the user didn't understand the answer they got so for example the bot may have answered correctly but for various reasons the the user basically didn't understand and you get into that under i don't understand loop so after say oh i don't understand or the user keeps on asking the question again and again it probably means that they don't understand the response so in that case you're really looking at an implicit transfer to an agent the other one which is actually very easy is where the user does it explicit explicitly part way through the conversation they simply say oh you're stupid i want to talk to a human so in which case you have the ability to transfer directly to the user as part of the conversation and the important thing is that the agent understands the conversation that the users had with the bot to that point because you don't want a situation where they've spent five ten minutes asking questions of the bot and then have the agent ask those same questions again so the agent has to be able to see the context of the conversation so they can add value rather than necessarily just repeat what the bot's already done so to answer your question where it comes where a human agent is needed is where the user didn't understand where the bot is not set up to answer that requirement or to the for from an emotional viewpoint uh some people just don't like talking to bots and really do want to talk to a real person and frequently that's that's the case with old age or older generation where they like talking to a real person so in that case rather than upset them allow them to talk to an agent and then do it directly yeah i have to say that's quite a common one for me when i when i'm using a bot quite often i'll get to a point and go right i need to talk to a human now i think the key thing as barry said is you get all the context so when the the human operator comes on it's not like they're coming in cold they get the conversation to that point uh okay there's a couple of other questions there so can the bot trigger the creation of an incident or does it have to be routed to an agent no absolutely so the bot can create the the case in the back end system and can also store the conversation so you can get a transcript of the users so even even if i didn't go to an agent i could have that conversation with the bot pushed into my call center and actually have a case and have the case id returned to the user in case they did ring back but yes you can actually take the transcript and have that stored in the case case system as well so the long answer is yes um what steps do i need to take to get started for chat or license yeah basically if you've already got the license that's great the just read this get in touch with us email um or uh or read chat to me on linkedin we can we can get you started in the renderer okay any any other questions all right i'm having a look i think we're good well thank you very much guys and yeah some good questions at the end there so um and thanks thanks very much barry for uh for joining us today that was uh really educated more than welcome said if you do have questions after the fact or they occur to you please do reach out to the folks at majia or your local oracle team but as i said um the digital assistant is state of the art this generation this routing this humanistic uh approach is very much um sort of the head of the curve if you like in in ai in the industry great cheers guys okay we'll give you back four minutes of your day absolutely real time thank you again thanks guys