Customer experience is centric to a company’s success. Yet bots are thrown in the front lines of customer interaction causing unnecessary friction with customers. Who wants to go through a phone tree? What if instead of bots, we improved customer support agent workflows? Anecdotally, keeping humans in the loop increases customer satisfaction. Ultimately augmentation, not automation, enables super-powered customer support agents, making everyone’s day more delightful.
In this presentation we’ll go through:
• Primer on Support Automation (SA)
• Overview of SA Market
• Approaches to SA
• Forethought SA Framework
• Where SA Excels
Want to learn more about Forethought?
Webinar Transcript
Adrienne McCrory: (00:22)
Welcome everybody who is joining as we get started, we’re looking forward to having you here and talking. I’m Adrienne, I’m the marketing manager here at Forethought, and we’ll give everybody one more minute before we get going. I hope you guys are having a great Friday. (silence)
Adrienne McCrory: (00:48)
Awesome. Okay. I’m going to share my screen. (silence).
Adrienne McCrory: (01:03)
Awesome. All right, I think I’m going to get going. It’s the top of the hour, actually it’s one minute after, so let’s go.
Adrienne McCrory: (01:13)
All right, I’m Adrienne McCrory as I was just saying, I am the product marketing manager here at Forethought, and I’m joined today with our CEO and co-founder, Deon. And he and I are going to be walking you through how to make AI a part of your life that you actually like. Probably Deon will be doing more of the talking than I will. But I just want to introduce some ideas here. Why are we here? Why does this matter right now? So just to give you a sense of what we’re going to be going through over the next 35 to 40 minutes, we’re going to be talking about why AI matters right now. Why it’s so crucial in this moment. We’re going to be talking about how you can better assess your customer questions and responses, and just the general flow of content coming into your organization. We’re also going to debunk some myths and we’re going to help you do an introspective look at what you need as a company and then we’ll wrap up.
Adrienne McCrory: (02:14)
So that’s our time. We hope it is super profitable for you. We know we’re all on Zoom so much these days and it is really important to have meaningful content and meaningful conversation, and that’s what we want to have with you guys today. So the fact of the matter is that AI is everywhere. We hear about it all the time. We know about it. You hear about it because of your work but also probably your life. And the fact of the matter for customer support is that those who use AI are going to get ahead and those who embed it into their workflows most strategically, the ones that really leverage its power are going to get further ahead. And so, we want to help you all address that today.
Adrienne McCrory: (02:59)
Without AI, I actually pulled this from Salesforce State Of Service Report, over 50% of agents feel like they are spending most of their time on mundane tasks if they’re not using AI. And we hear lots of different things about why AI is meaningful to agents, to customer support, to customers themselves and we know that it’s time to use it. The problem is that too frequently, what we see and why AI sometimes still feels like a buzzword is that we get misdiagnosed. As customers support lead yourselves, working in operations, wherever you are in the organization, you might be getting a misdiagnosis of what kind of AI you actually need that will actually help move the dial and help your customers get what they need faster. And we want to unpack that today. Why is that happening and what can we do to make that experience better?
Adrienne McCrory: (04:02)
What happens when you get a misdiagnosis is that you have a bad experience. Maybe you’ve gotten burnt by AI before. Maybe you’ve had a dud AI that just didn’t do what you wanted it to do. Maybe it just was a total mismatch. And then what also ends up happening is you have your unresolved issues. You still have customers who need certain things from your organization and you’re not able to provide it, and so that’s what we really want to avoid.
Adrienne McCrory: (04:34)
So how can we ensure a better relationship? That’s really where Deon is going to come in and share some insight for us today to talk us through this. Deon, like I said, is our CEO and co-founder, he is also an ML and AI expert. He’s a trained software engineer. He understands customer support systems because he has been talking to customer support leaders for several years now. He also has stocked shelves before in his life. I think understanding being in the nitty gritty of this thing is super important, because at the end of the day it’s about your customers, right? And he also is a parent of two children. So I added those last two, he did not ask me to add those last two. So Deon, I’ll let you take it away.
Deon Nicholas: (05:23)
Awesome. Thank you so much, Adrienne for the great introduction. Thank you everyone for joining. As Adrian mentioned, I’m Deon Nicholas, CEO and co-founder of Forethought. I started this company three years ago on a mission to use artificial intelligence to enable your inner genius. And we started by building artificial intelligence for customer support teams. So as you can tell, we’ve spent a lot of time thinking about AI, thinking about what it really means, what it can deliver and what it can deliver today and what it can deliver tomorrow. And I spent a lot of time with customer support leaders from many different industries whether that’s high tech, E-commerce and so on. And we’ve been seeing a time and time again that AI in many ways gets a bad rap, in part because though it has a lot of potential and promise, there’s been a lot of, as Adrian mentioned, misdiagnosis of the issues that AI can help solve.
Deon Nicholas: (06:19)
So today I’d love to talk a little bit about how we can go about making AI work for you. So the first question that we often get is, why AI? Why now? And the real thing to think about in 2021 is that customer support fundamentally changed in 2020. Besides global, civil unrest, a lot of things happening, we were hit with probably one of the largest, most dangerous pandemics of our lifetime. And COVID meant a lot of things in particular to the business world.
Deon Nicholas: (06:52)
So first and foremost, COVID forced a remote work style. I don’t know if a lot of folks are following the news but obviously there were lockdowns in and around March of 2020 across the world. And very importantly for our conversation today, in countries like Manila, Philippines where the largest population of business process outsourcing companies are, the folks who supply customer support agents and tools, they shut down completely in and around March. And what this meant was that a lot of companies, whether you were in Manila, whether you were in the United States, or Europe, or South America saw their customer support teams literally go offline. Agents were being sent home. Agents didn’t have equipment to work from home and so you saw a complete loss of productivity across many, many, many industries, particularly for customer support. I don’t need to tell you that how remote life has changed all of our lives.
Deon Nicholas: (07:58)
Secondly, the thing that happened with customer support is that due to all of the uncertainty going on in 2020, we saw a simultaneous exponential growth in customer need. So for many companies, for many businesses, overnight you saw a spike in 5X to 10X volume in customer support inquiries. Imagine, for example you’re an airline, the number of canceled flights, the number of refunds, the number of things that had to happen when this shift happened, when lockdowns happened, when remote work started happening. Or imagine being a bank with PPP loans being rolled out and things like that, how many different inquiries you got?
Deon Nicholas: (08:40)
What we were seeing is both with our customers and with experts around the industry, we were seeing that volumes were going through the roof across all industries, and so you had the forced remote work lifestyle. You had the exponential growth and customer need, but at the same time, in the midst of all this, customers like you and me still desire a good customer experience, still desire a positive relationship with the people we buy from. And so what this meant is that the need for artificial intelligence to both help with the productivity, to help with the demand, but also to continue to alleviate stress, improve delight, and create a great customer experience has grown exponentially.
Deon Nicholas: (09:26)
And so, these are the reasons why today it is very important for you if you are a business leader to invest in artificial intelligence in to help with these problems. But again, what we’re finding is that one of the biggest problems is that people tend to misdiagnose themselves, misdiagnose AI. There’s a ton of different AI out there, but every single organization has different needs. And this is one of the things I like to stress time and time again because when we see customers coming to us and they say, “Hey, we want an AI can you help us?” The first thing you need to do is actually start asking questions about your organization. What is your business model? What kinds of customers are you supporting? Do you have high volumes? Do you have complex tickets? We’ll talk a little bit about that more in the next little bit, but the point is that no two businesses are alike.
Deon Nicholas: (10:17)
And that brings me to my second point, is that not all AI is created equal. No two AI vendors or solutions actually do the same thing. And when you dig under the hood at how all of these works, there’s multiple different kinds of AI that you can apply. And so, the first thing and the thing we’re going to be talking about today is how to go and assess that. Okay, so the first step in figuring out how to get a true diagnosis of what you need is to assess your needs.
Deon Nicholas: (10:47)
Step one, figure out what are your customers needs? What kinds of questions are they asking every single day as well as what kinds of responses are they looking for? And so we’ll talk a little bit more and we’ll actually spend a lot of the time here right now talking about how to diagnose customers needs. And we’ll be walking through a quadrant shortly on allowing you to figure out what that means. And then the second step after this if we go back to the previous slide, we will talk later on in this webinar about how to assess your organization’s need. So what are the resources that you have? How much work do you want to put in? What do you think about implementation and other things that will enable you to make a good decision about which AI to use.
Deon Nicholas: (11:33)
Okay, so jumping in, let’s talk about what I like to call the assessing customer needs quadrant. So on your screen you can see a four by four quadrant here. And the way we like to think about customer support and businesses here at Forethought again is not one size fits all, but really tailored to what kind of business you are. On the left end of the spectrum, you see simple questions. So the first thing you need to ask yourself is, how complex are the questions that my customers are asking me? On the left hand side is simple questions and on the right hand side you can start to see more complex questions, and we’ll into that further in a couple seconds here. And then the other dimension and access that you need to think about is, what kind of responses are my customers expecting from me? On the bottom side, it is action based responses. They want you to do something or help them do something. And on the top side it’s actually knowledge and information based responses. They have a question, they want a piece of information that they want the answers to.
Deon Nicholas: (12:37)
And so, let me talk a little bit about all of these four quadrants and then we’ll jump into each one individually and what that really means for you and your organization as you’re doing the assessment. So overall when you actually take these four quadrants, sorry, we’ll go back the previous slide. When you actually think about these four quadrants, when you mix simple versus complex questions, as well as action versus knowledge oriented questions, you get four different kinds of customer support need.
Deon Nicholas: (13:10)
So on the top left, it is the, tell me, types of questions. They are simple questions and they need a knowledge response, and we’ll talk a little bit more about what that means, but the need there is to make the experience delightful. In the bottom left is the, do for me, types of responses. Your customers are asking very simple questions but they want you to go and take an action and perform an action on their behalf. And these are the, execute task for me, type issues. On the top right we have the more complex knowledge oriented questions where your customers are looking for you to teach them something. And then on the bottom right, you have the most complex kinds of questions which are complex action oriented questions, or your customers want you to help them accomplish an investigation or a really difficult task. And I know this is a lot, so we will slow it down and start talking about each of these quadrants individually.
Deon Nicholas: (14:05)
And so as we go through these, I want everyone on the call to think about, as a business, what do I need? What kind of organization am I? And I’ll start to talk about what these look like. So the first quadrant is simple knowledge based responses, so simple questions, knowledge based responses. These are the questions where your customers say, “Please tell me something.” So imagine you are, say a B2C company or possibly a B2B2C company where you’re selling to small businesses who have lots of customers and typically you have a strong virtual presence or digital offering. So we have some logos here, some examples of… Not all of these are Forethought customers, but in this particular case, here are some customers and or companies that we’ve seen and the kinds of support we tend to see here that fits into this, tell me, type of workflow.
Deon Nicholas: (14:57)
So these are the kinds of questions that for example, customers might ask who, what, where, or when questions. They’re inquiring for a specific piece of information and they need a little nudge towards the right content to be successful. So an example of this, if you’ve heard of Masterclass for example, they’re an online education company, very B2C and they have a strong virtual presence because their entire product is online. Customers can ask questions like… Imagine asking a question like, “Hey, can I meet the instructor?” These are very simple question with a knowledge oriented response. Maybe the answer is yes. Maybe the answer is no. Maybe the answer is, here are the cases in which you do it. But the really big thing to think about here is these are the kinds of questions that require either redirecting or pointing to a knowledge base or answering with a simple piece of information.
Deon Nicholas: (15:47)
Other examples can be Instacart, imagine, “Hey, where’s my order?” Things like that. And we’ll talk later about what kind of AI and what kind of automation or needs this might map to. So these are the simple, tell me, types of questions and you’ll often find you fit into this category if you’re a B2C or a B2B2C company with a strong virtual presence.
Deon Nicholas: (16:10)
The next quadrant is what we call the do for me kinds of questions. So still on the simple end but the customer’s not just asking for a piece of information, a who, what, where or when question, but they’re actually asking you to do something for them. This often occurs in B2C companies of an E-commerce or brick and mortar nature. So if you have a heavy sales motion or process where you’re actually selling physical products or digital goods, you often fall into this bucket. Your customers are asking you to perform tasks on their behalf. Examples can be, “Hey, can you issue me a refund? Can you reset my password? Can you change my flight ticket,” and things like that. So you’re actually selling a physical service or something that is complex enough that the customer will inquire about it and require you to actually perform an action on their behalf. You’ll find that many of the tasks here end up being repetitive but require action nonetheless. We’ll talk a little bit more about what that means and how AI can actually help you solve those.
Deon Nicholas: (17:17)
In the next quadrant we have the, teach me, kinds of questions. And so these are knowledge oriented but you’re now starting to get to the more complex problems. This can often happen if you’re a B2B or B2B2C business, often with high tech or a highly regulated industries such as FinTech, education, and other things. Benefits, HR, software, those kinds of software. What ends up happening here is that the kinds of questions your customers are asking end up being of the how and why variety. So we’ve moved away from the who, what, where, and when, where they just need a simple factual answer. And we’ve now moved into the how’s, and why’s. These are questions that require and where your customer expects explanations on the how, on the why behind something like this.
Deon Nicholas: (18:11)
So this might be your inquiring of your recent stock trades at a FinTech company. And you’re asking questions about how and why things are the way they are. Or this might be, you’re asking your education technology company questions about courses or things like that. Or you may have an issue with your video provider and you’re asking questions about, “How can I install something? Can you walk me through this?” And so these questions typically require a little bit more nuance, they often require calculations and they often require domain expertise in order to answer them.
Deon Nicholas: (18:49)
And then last but certainly not least is the complex action oriented questions. These are the questions that require the biggest lift in an organization. And again you may fit into one of these quadrants as a business as you think through how we’re talking here. Most companies typically have a predominant style. You can think of it almost as a personality type or any Agram type so to speak. But most companies have a predominant style of what their customers are asking, but you will often find some percentage of your tickets and your questions can be in others. In this case, we have the complex action oriented tickets, where your customer is asking you to help them do something. This is often a B2B or SaaS product as described with heavy duty creative capabilities.
Deon Nicholas: (19:36)
So this is products with analytics, or surveys, or designing, or flow charts where your customers are trying to get something done that is really complex. And by the time they come to you for a question, for a support ticket, they’ve often tried most of their own avenues. And when they get to you, they actually do need help. They need help potentially logging into their system not just because they forgot their password but because they need help actually going and creating a new flow chart or using the tool to its fullest capabilities. And you’ll often find your most engaged customer support queries come from your most highly active power users.
Deon Nicholas: (20:17)
So I know that’s a lot to think about, but the key takeaway there that I want to re-emphasize is that the biggest mistake we see that customer support teams and any other team looking to leverage AI does is they think about AI as a one size fits all. Hopefully everyone on this call today, as you were hearing me talk through the different quadrants, you could hear versions of the business where you’re like, “Yeah, that actually sounds more like me.” Or, “This other business really doesn’t sound like me at all.” And that’s really the first and most important thing that you want to get out. Once you’ve identified the quadrant where the bulk of your tickets live, now we can talk about which AI makes sense for you. And as a note, we will be giving this out as a takeaway, as a handout for anyone here at the webinar today and will be available online for future use, so we’ll send all of this out.
Deon Nicholas: (21:11)
But let’s just walk through it really quickly. If you are in the top left quadrant. So again, these are the simple knowledge oriented questions where the goal is to tell me something and the end goal is to make the experience more delightful. There are a few different ways you can leverage AI, and leverage automation, and leverage intelligence to make this work. The most important thing in this quadrant is to remember that these are actually the questions that possibly could have been solved with self-service. These are the questions that possibly didn’t need an agent at all, possibly could have been solved on your website with better content.
Deon Nicholas: (21:53)
So the actual first thing that you need to think about when you are in the top left quadrant is, what is my content doing for me? And so, you can often start by evaluating your own self-service content, that means your public knowledge based articles. Things within if you’re an app company or a software company, things within the product. And then employing artificial intelligence to either help you detect knowledge based gaps. So to figure out where questions are coming that you maybe don’t have good knowledge content for or to improve site search. And so that means enabling discoverability of your content before people even need to submit a ticket at all.
Deon Nicholas: (22:37)
Another way that you can use artificial intelligence is through chatbots. Here, these kinds of chatbots are the ones that we call deflection based. So any chat bot that can respond automatically with content, with macros, with other things like that or other types of ticket deflection software can be really helpful here for these simple questions that don’t actually need to go into an agent at all. And then the last is macro and template automation. We actually do a lot of this here at Forethought with our Agatha Solve product, but macro template automation is when you actually take responses, canned responses, macros, past responses that your agents have used and you use AI to actually bring those to the forefront and automatically reply to that. And then again as mentioned, wrapping it all together by using artificial intelligence to understand knowledge base gaps so that you can improve your macros, improve your self-service content and improve your templates.
Deon Nicholas: (23:32)
So that’s the top left quadrant, and note that these are different vendors. With site search, there are some vendors that do just site search. There’re some vendors that just do chatbots. There’re some vendors who do macro and template automation and there are some vendors who do knowledge base gaps. Obviously, if you’re working with a Forethought, we are a full service AI platform and try to actually address these particular issues, but it’s just something to keep in mind that not all AI is created equal.
Deon Nicholas: (24:01)
In the bottom left quadrant, again you have these simple questions but now they’re requiring an action. “Please go issue me a refund. Please go reset my password. Change my flight.” The thing to know about this quadrant, again because they’re actually the simpler questions, it’s not necessarily just a matter of self-service content, but what really here is going on is that there’s often a gap in the product itself. If people have to go and submit a support ticket in order to reset their password, then that’s often a sign that you or the product team can actually go and implement a workflow within the product to go and reset your password.
Deon Nicholas: (24:40)
And so what you’ll find here is that the kinds of AI you will use if you’re looking at these simple action based, do for me questions is, one, either automations or chatbots that specialize in automations to go and perform these actions or you’re looking at rules based or action based decision trees in order to hard code workflows. This could often work well, it’s not actually a form of AI, but can work well for the very simple based responses. And it does happen to be the kind of thing that people mistake for AI a lot which is why I’m actually mentioning it here. But oftentimes, this can even be something as simple as triggers, and rules, and so on.
Deon Nicholas: (25:20)
And then lastly in this bottom left quadrant, identifying trends, identifying product gaps again that you bring information back to your team in order to make better decisions. Top left quadrant as we think through the AI here, knowledge based responses and complex questions. Here, the main issue is that the problems themselves actually do require a bit of nuance and they require expert agents. So the types of AI you’re going to use here are triage and routing to experts to make sure the right agents are working on the right problems. You’re going to have AI to catch complicated issues and you’re going to have AI to assist those agents with the problems. This can often be in the form of AI knowledge bases, enterprise technology, and other technology like that. And then lastly detecting knowledge based gaps where we can actually figure out what issues should have actually been self-served so that your agents can focus more on those more complicated problems.
Deon Nicholas: (26:24)
And then the bottom right quadrant in the, do for me quadrant, complex questions, action oriented tickets is agent assistant and ramp up tools. The point here is that these truly require experts, these truly require domain expertise. So the only thing, or one of the main things that you can do here is help get your early agents to behave like experts. And so agent assist tools, ramping up, helping your agents get more up to speed, knowledge capture so that you can take what the best agents are doing and then teach that to your earlier agents, identifying trends. There’s often a ton of integrations as well as robotic process automation which is ways again to manually create these more complex workflows. I’m going to pause there and I think we’re halfway through the time. So Adrienne, I just wanted to hand that back over to you.
Adrienne McCrory: (27:18)
Awesome. Thank you, Deon. Yes, let’s take a brief break. I feel like Deon might need a drink of water after that talking. So we’re actually going to take a break from both of us and hear from some other Forethinkers at Forethought, and hear about two myths that we have heard a few different times. So in order to do this, I’m going to re-share a video, just give me one second. And this hopefully will shed some light on different things we’ve heard. And maybe you’ve heard these too.
Irene Shao: (28:05)
Hi folks. My name is Irene and I am a product manager here at Forethought. And today I want to quickly address this common misconception that AI is only useful for deflecting customer issues. I mean yes, AI is often useful for deflection. When a customer ticket comes in, AI could help answer it before the ticket ever gets to an agent. But what about all those tickets that do land in front of an agent? Can AI help with those? Absolutely.
Irene Shao: (28:33)
Now, imagine you’re a new agent and you have no idea how to answer the ticket in front of you but your AI helper shows you a few responses that more experienced agents had already written to answer a very similar or ticket. Or if you have a ticket asking the exact same question that you addressed last week, instead of searching through all your past tickets for that one answer so you can copy and paste, imagine your AI helper surfacing all relevant past cases that you answered in the past. Or you know exactly which article you need to send back to the customer, but instead of having to look through the entire knowledge base your AI helper pins that exact article next to your reply box. All that is to say, as you think about AI, consider the ways in which it can help you even beyond deflection.
Emma: (29:23)
Hi everyone. I’m Emma.
Stephen: (29:24)
And I’m Stephen. We’re account executives here at Forethought.
Emma: (29:27)
One of the common concerns we hear from our customers wanting to leverage AI tools is that they don’t have the time or the resources to get AI up and running in their org.
Stephen: (29:35)
This is understandable given the market for AI tools. There are indeed many tools that require a big lift upfront from your team in ongoing maintenance, setting up knowledge trees, keyword triggers, et cetera can be time consuming and costly.
Emma: (29:47)
But it doesn’t have to be that way. At places like Forethought, our stellar team of engineers builds custom models for your company. This means that we’re training on your historical data which requires no build out on your end. Training the models to understand your workflow actually only takes a few days.
Stephen: (30:04)
We often see measurable impact within just two weeks of launching our AI. With one of our customers, we were deflecting 24% of their tickets within the first month. Many other tools out there take three to six months to see similar results.
Emma: (30:17)
In a nutshell, some AI does truly take forever to implement and requires a ton of work from you, but some AI is sophisticated enough to take the burden off of you entirely.
Adrienne McCrory: (30:28)
Awesome. Okay, commercial break is over. We can get back into the content and I’m going to go back to our presentation. If you guys have questions as we’re chatting, please post them in the chat, we’re happy to take them. Even if we can’t address them right now, it’s just super helpful to see what things you guys would like more clarity on. We know it’s a lot that we’re going through and we’re always happy to chat with you at a different time, one on one, to talk more about any of the things that we’re talking about.
Adrienne McCrory: (31:02)
Awesome. Okay, let’s get back into it. So we looked at the one side which is doing a diagnosis of what your customers actually need and now we’re going to look at internal issues. So I’ll turn it back over to you, Deon.
Deon Nicholas: (31:18)
Awesome. Thank you, Adrienne. And thanks to everyone, it’s fun to see some of those myths busted. AI is definitely one of these industries where there’s a lot of buzzwords floating around, right? And so it’s great to see kind of peeling under the hood. When is it real AI? What does it mean? And hopefully this has been useful for you all. As Adrienne mentioned, if you have any questions for us, feel free to chat to the panelists. We’re on Zoom here, so we can definitely respond to any of those. We also have the Q&A feature. If you have a question that you want for the whole group, and we’ll pause and answer those as best we can.
Deon Nicholas: (32:00)
All right, so moving forward, assessing your internal needs. We’ve already talked a little bit about just understanding the flow and the flavor your customer support personality, so to speak, of your customer’s questions as well as your responses, now it’s time to assess your internal needs. For example, one, think through how much renovation in your systems that you want to commit to. There are some technologies, artificial intelligence, or other wise that require a heavy lift. And then some that don’t. Part two, you should do an internal audit of what data does your organization actually use. What data is are your agents actually using in order to resolve questions and this will help give you a sense of what kinds of AI you should actually be using. And then third, the biggest tip is literally, don’t try to solve all of your problems at once, but rather start with the greatest need and build on that. So let’s quickly elaborate on all of these and then we’ll stop for questions and I think we’ll be good to go.
Deon Nicholas: (33:01)
So step one, think about how much renovation do you want to do. There’s actually a spectrum, and some is better than the others, and some really just depends on what you need as an organization. On the far left hand side, and this is actually where that myth number two came from, is that most organizations think you need to hire somebody to implement a new system. And this can be true, it really depends on your needs and really depends on the vendor you go with. There are some vendors that require a heavy lift, require a lot of manual input in order to onboard the system as well as ongoing maintenance for example, the content for the system itself and so on and so forth. So on the far end of the spectrum we have seen there are some AI vendors and some companies who actually go the route of hiring a program manager or someone to work on the actual AI system.
Deon Nicholas: (33:54)
In the middle is, you can actually get away with building out new content to fit the new AI framework. So this means you don’t necessarily need to implement a completely new system, but the data that you’re indexing or the data that you’re using such as your content should actually be maintained. So you might have a content writer on your team. If, for example, you think about these self-service articles, or macros, or you might have a writer it internally to work on internal documentation, and so that that is kind of the medium lift approach. And then there are vendors for example, like Forethought where you can use your existing content and identify the best things that can actually be applied to this AI. So there are some AI out there where, if you’re looking for a quick implementation and you don’t want to have a person overseeing that, then you can actually use your existing content, use your existing knowledge bases and integrate with that AI.
Deon Nicholas: (34:57)
Part two, which data does your team actually use day to day? And this is one actually trips up a lot of folks. On the far end of the spectrum is fully baked customer-facing material. So again, we talked about your knowledge base on your public website, or maybe it is your website, or your app, or things available in your app or at your storefront. These are things, if your team is using this data, then you’ll often have AI that ends up either integrating with that or replicating that.
Deon Nicholas: (35:27)
In the middle end of the spectrum is internal documents. So these are still knowledge bases but they’re often things that your agents are using. And then on the far right end of the spectrum is internal databases customer info. So these are things like, when you need to check account status, or when you need to actually log into customer systems in order to get work done. And so, they’re actually varying things to think about. When onboarding a new AI vendor, before even going to think about deploying new AI, it’s often good to perform this internal audit, so to speak, of where do my agents actually get data from when they’re resolving questions.
Deon Nicholas: (36:05)
And then lastly, think about what is your greatest need day to day. So again, as you thought about the quadrant, as you think about what business you are, your customer support personality. Where is that greatest need? You’re going to find some technology in artificial intelligence or automation that is useful for deflecting repetitive tickets. You’re going to find that maybe we just need to improve our SLAs. It’s not just about reducing volume, but it’s about reducing that first response to so that we get back to our customers more quickly. Thinking through that, maybe it’s about organizing our tickets so that again those experts can get their hands on the right tickets at the right time. Or maybe it’s about finding the biggest issues such as product gaps or knowledge gaps. Or maybe it’s about pulling all of your knowledge together. So it’s really, really important before you go and deploy an AI vendor to actually think through, what are your organization’s big OKRs, objectives and key results, so to speak, or key performance indicators? And that will actually be very different for different people.
Deon Nicholas: (37:07)
So once you’ve gone through and done those three steps, so that is one, thinking about how much renovation you need documenting that. Two, thinking about which data your team actually uses day to day. And three, assessing what your greatest need is. You can actually be in a very, very strong spot to put together a request for proposals on RFP and start thinking about how you’re going to sift through and weed through the different kinds of AI, so to speak.
Deon Nicholas: (37:39)
Now last but certainly not least, thank you all for coming. AI is here, it’s a part of the future. Don’t get left in the dust as we like to say. But hopefully, based on this webinar you feel armed with information that’ll enable you to really think deeply about how to use AI for your specific use case. If there’s one thing that I want you all to come away with here, it’s that not all AI is created equal and the AI that will work for me may not work for that other organization. And so if you have any questions, don’t hesitate to reach out to myself, or Adrienne, or anyone here at Forethought with questions. And the other thing I will say is that we will be hosting future webinars. Our next one, thinking about how you can actually avoid getting duped into high maintenance AI over time. I’m going to hand it back over to Adrienne. Thank you all so much.
Adrienne McCrory: (38:31)
Awesome. Thank you, Deon. Thank you everybody for your time today. We know we packed a lot in here, and there’s a lot to think about, and we set you up to go do some introspection, so we hope that you will go and do these things within your organization. I will be sending this document that you’re seeing on the screen right now out to all of you so you can go out armed with some of the things we talked about today, including the quadrant, which I think is super helpful for thinking about just the landscape of AI as well as some reflective questions for yourself internally. And then as Deon very nicely went through and described the different types of AI that you can use if you’re in one of these quadrants, you have this too.
Adrienne McCrory: (39:17)
So I just want to point out one quick thing, I know Deon said this. But you might fall squarely into one of these quadrants, you might be bleeding into one of the edges of a different quadrant, it’s definitely not supposed to be oversimplifying, but we are trying to make sense of AI with you and for you. That’s all. Like Deon said, reach out if you have questions. We’d love to hear from you and thank you so much for your time today. Have a great rest of your Friday.
