Latest / CXOTalk / Will AI Change Customer Service Forever in 2025? | CXOTalk #879
Transcript
- 0:00AI is reshaping customer communications with terms like
- 0:04virtual workforce and AI receptionist.
- 0:08I'm Michael Krigsman, and I spoke with Damon Covey from Go
- 0:11to to explore how small companies can use AI agents to
- 0:16improve both efficiency and customer service.
- 0:21One of the main customer services challenges is being
- 0:23able to communicate effectively with your customer over the
- 0:27channel and the way they want to talk to you.
- 0:30If a customer wants to text you, if they want to do reach out to
- 0:33you via Instagram or Facebook, things of that nature, being
- 0:36able to communicate with them that way.
- 0:38And traditionally what that looked like for customers is
- 0:41they got to go use all those different platforms to
- 0:42communicate with those customers and they didn't have a way to do
- 0:46that. Now there are solutions that
- 0:47bring all of those things together.
- 0:49So it makes both the employee experience or the business
- 0:53experience and then also the end user experience as seamless as
- 0:57possible. So as customers needs to evolve,
- 0:59we've added more and more channels and opportunities to
- 1:01communicate over the years to to that cloud communication
- 1:04solution or in the business of providing the opportunity to
- 1:07talk to customers, to communicate with customers,
- 1:10schedule appointments, things of that nature in the way that
- 1:12customers want to talk with the company directly and that's it.
- 1:16That's a challenge traditionally, especially for
- 1:18smaller customers. Can you give us a few examples?
- 1:20If you have a dentist practice and someone wants to communicate
- 1:24with you and set an appointment, they're very often just going to
- 1:27be doing whatever they do during the day and and a text is much
- 1:30faster to set that appointment. So that's one example.
- 1:33Another example I can give you from my personal experience is I
- 1:35had a pipe that burst in my basement.
- 1:37I was trying to reach out and find a plumber.
- 1:39I didn't know a plumber. So I started calling down the
- 1:41list of plumbers and the first one to text me back got my
- 1:45business right. And so being able to communicate
- 1:47again with the customers in the way they want to, the way
- 1:49they're communicating in flow is a really critical part of
- 1:52especially a small business practice.
- 1:54So making it easier for customers to interact with the
- 1:58business. Making it easier and then also
- 2:01having it be integrated back into the systems they care
- 2:04about, their CRM or other systems that they might be
- 2:06working with, keeping them in those systems to make sure
- 2:09they're tracking that information as well.
- 2:11So the integration with their existing systems, their existing
- 2:16processes is essential here. It's critical to make sure that
- 2:21they can continue to do business the way they want to, but also
- 2:24communicate with their customers in the way the customers want to
- 2:26communicate with them. How can AI make this process
- 2:31easier, better, faster? AI has the opportunity to really
- 2:36take, especially if you look at a small business, they are most
- 2:41of the time multitasking, right? They're dealing with customers.
- 2:44They're also dealing with things inside of that business.
- 2:47And so they don't always have the opportunity to answer the
- 2:50calls or answer the key questions, things of that
- 2:52nature. They might actually lose some of
- 2:54those customers that are calling in.
- 2:56AI has the ability to be a force multiplier and answer the phones
- 3:00as an example 24/7 for them, answer specific questions and
- 3:04then transfer that information along with the caller when
- 3:07appropriate to the right person who can help them and get that
- 3:10to be a more efficient rocess. Damon, I've heard you use the
- 3:13term virtual workforce. What?
- 3:17Is that virtual workforce is essentially a way to harness AI
- 3:21technology to do things that you would like a human be able to do
- 3:25what they may not be available. If you think about humans in
- 3:28your environment, they're not only answering calls and texts
- 3:31and things of that nature, they're doing things also.
- 3:33So you want to automate those repetitive tasks.
- 3:35And so AI can take a lot of that over, can start to automate
- 3:39that, whether it's answering the phone in some cases we talked
- 3:41about whether it's possibly answering questions about
- 3:44locations, hours, things of that nature.
- 3:47So really use AI in a very smart and practical way to get into it
- 3:52and automate those tasks. That's where you start to see
- 3:54the AI virtual workforce emerge. And we'll see more of that as we
- 3:58go into the future as well. We hear a lot about AI agents
- 4:03and agentic AI. Is a virtual workforce the same
- 4:08thing? It's a part of it.
- 4:10So if you think about virtual workforce and agents, that's the
- 4:14tip of the spear for answering calls, automating mundane tasks,
- 4:19things of that nature. And what Agentic does is takes
- 4:21that information and takes action on that, right?
- 4:24So if you need something to happen in somewhat of an
- 4:28autonomous fashion, that's where a gentic AI can come into play.
- 4:31So for example, if you want to pull something out of ACRM and
- 4:35then insert that back as you're talking to a caller, that would
- 4:38be an example of a scientific AI on the front end.
- 4:41And then it will get more sophisticated as we go forward
- 4:44and and do more sophisticated options and advanced actions as
- 4:46well. And so how is that different or
- 4:49the same from the virtual workforce?
- 4:52It's part and parcel to the same thing.
- 4:53Virtual workforces will just mature over time.
- 4:57So to be able to again, automate those tip of the spear tasks and
- 5:00then also to automate other tasks that are smarter, right,
- 5:04that require connecting to multiple different systems,
- 5:07require, you know, multiple different pieces of input.
- 5:09That's where a gentic AI really has power is in connecting those
- 5:13things together and automating A workflow to satisfy a particular
- 5:17problem that might dip in and out of five or six systems, as
- 5:20an example. And then how does the virtual
- 5:23workforce relate to that workflow that you just
- 5:27mentioned? Virtual workforce will be at the
- 5:29front end of that. And so an example of that is a
- 5:33receptionist be able to answer the call, understand the intent
- 5:36of the caller, answer some simple questions, and then
- 5:39potentially transfer that person over in the future.
- 5:41We see that getting more and more sophisticated, whether you
- 5:44can actually take actions as opposed to just answering the
- 5:47call. That's how the two fit together.
- 5:49You've just released an AI receptionist.
- 5:53Tell us how that works. Air receptions is something that
- 5:56anybody can add into their telephony system that will
- 6:00answer phone calls for you. If you think about it, Michael,
- 6:03the most challenging part of AI as as it relates to customer
- 6:07communications is the voice section, right?
- 6:09So that's what we took that one on. 1st.
- 6:11What this allows you to do is put that at the front end or
- 6:14frankly at the middle or even at the end of your telephone call
- 6:18and answer that, answer that call, answer questions and
- 6:21transfer that over to the appropriate person on your side
- 6:25with the intent also summarized. So imagine, you know, before
- 6:28you're just getting a call from someone out of the blue, you
- 6:30don't know what they're calling about.
- 6:31We can capture that information and transfer that over to
- 6:34someone where they can actually understand what this person's
- 6:36calling about. Are they upset or not?
- 6:38Not what they're actually looking for.
- 6:39And you can automatically know how to come out of the call and
- 6:42how to better handle that in that call.
- 6:44So the virtual receptionist interacts with the caller and
- 6:49feeds information back upstream to a person who may or may not
- 6:57interact with that caller. Is this correct?
- 6:59Yes, in many cases that's correct.
- 7:01And we can also extend that a receptionist to answer again,
- 7:05common questions, right? So if you want to help that AI
- 7:09learn over time by feeding it information that it should
- 7:12answer as you start to observe what's happening and start to
- 7:15observe that. But yes, ultimately it probably
- 7:18ends in some human interaction. And the idea is to augment the
- 7:21human interaction with the caller and make that at
- 7:24resolution as speedy as possible.
- 7:26So to some degree, the virtual workforce is an autonomous
- 7:33process that interacts with the caller.
- 7:37Absolutely, on its own. It will, with your permission of
- 7:41course, answer those calls, answer those questions.
- 7:44But it can be, and it does operate within its sphere of
- 7:47influence, autonomous from someone else actually having to
- 7:51manually take that call or take an action.
- 7:53And so there's a set of boundaries that are built in.
- 7:56You mentioned scope or sphere of action.
- 7:58Correct. That's a really important part
- 8:00of this actually, Michael, is what we don't want to do is just
- 8:03wire an AI receptionist to an LLL, right?
- 8:06That would be a bad idea because it might actually give some bad
- 8:09information or something that you weren't expecting it to.
- 8:13So making sure you tune the air receptionist to give the proper
- 8:17information only with the guardrails you wanted to and
- 8:20starting with a very scoped set of of responsibilities and
- 8:24building up from there is is the way to to help it learn with
- 8:27your system. So the set of boundaries is
- 8:31vitally important here. Absolutely.
- 8:33It's probably the most critical part of this to make sure.
- 8:35And again, what I always tell people is or something
- 8:37practical, start with something that's a simple problem that's
- 8:41repeatable, that you can solve over and over again that you
- 8:43feel good about an AI receptionist talking about.
- 8:47And then build up from there as you start to see how those
- 8:50reactions are going as you go forward.
- 8:52Damon, what benefits does a virtual workforce bring to go
- 8:56To's customers? So it starts with really the
- 8:58ability to have any interaction handled 24 by 7.
- 9:03So if you've got your employees that are doing something else
- 9:06and they can't answer the call, that's number one.
- 9:08The other thing is ensuring that you can free those employees up
- 9:12to do higher value things, more beneficial things and look at
- 9:16other opportunities within within your environment.
- 9:18So if they're doing the same thing over and over again, now
- 9:21we can free them up to do that. And then I would say the other
- 9:23thing is really just making sure that because you're answering
- 9:26swiftly and giving them feedback right away, really enhancing
- 9:29customer satisfaction from the end user perspective, they're
- 9:33getting information they need faster than they would have
- 9:34previously. This notion of improving
- 9:38customer service for the caller is really important.
- 9:41Can you elaborate on that? With an AI receptionist or or
- 9:44AI, you have the ability to guarantee the response that the,
- 9:49that the user or the caller in this case is getting.
- 9:52And so that gives them the ability to automatically have
- 9:55their call answered at all times.
- 9:57So that that's number one is, is a primary benefit.
- 10:00And then to get a guaranteed answer.
- 10:01And what that is and if they need to be have a guaranteed way
- 10:05of getting interaction. So those are all three things
- 10:08that customers are looking for that immediate response.
- 10:11They want that immediate interaction.
- 10:13People don't wait anymore to get feedback from companies they
- 10:16want to do business with, and this gives them that immediate
- 10:18feedback. How is this different or more
- 10:21beneficial than traditional customer service channels?
- 10:24Traditional customer service channels always need a human
- 10:27intervention and this is not meant to replace that human
- 10:29intervention. What this does is helps you
- 10:31scale your business, right? And so that's where it's special
- 10:34is again, if you're a, if you're a small customer, part of what
- 10:38you're doing is just, you know, the business of making sure that
- 10:40your business is done and scaling it, making sure you can
- 10:43answer all those calls, making sure those they were having to
- 10:46have an agent behind every single call.
- 10:48And again, freeing up those agents to do more, higher
- 10:51interaction, better sales process, better service process,
- 10:55things of that nature. That's really what this is about
- 10:57is augmenting your human workforce so that they can do
- 11:00things that are higher value again to the to the company on
- 11:02the back end. So there is an efficiency
- 11:04benefit to the go to customer and a service benefit to the end
- 11:12user that's calling in. Correct, You definitely get a
- 11:16benefit both ways. In terms of accuracy, in terms
- 11:19of of how fast the call is answered, in terms of how fast
- 11:22it's responded to, which will by itself gain you a higher
- 11:26satisfaction rate. Damon, let's talk about
- 11:28deployment. What is involved with
- 11:31implementing or deploying this kind of system?
- 11:33That's a critical decision factor.
- 11:35Small customers have not had the ability to, you know, capture
- 11:39something off the shelf, like implement that on their own.
- 11:41So the AI receptionist, as an example, introduces the ability
- 11:45to have a small and simple operation that you can feed
- 11:48things like K BS knowledge base articles as an example.
- 11:52We can feed FAQs into, right? We can start to actually put
- 11:55questions in there, things of that nature.
- 11:57So it makes it very, very easy and practical to get up and
- 12:01going off the ground. And then I think the other thing
- 12:03that's really important about this, Michael, is you can
- 12:06observe the system over time as well.
- 12:08So that's a critical differentiator also is it's,
- 12:11yes, it's easy to get up and running, but can I see how it's
- 12:13actually doing? Can I, can I see how the
- 12:16customers are responding to it? Can I listen to the calls?
- 12:18All the above is there. And so you can tune that as you
- 12:21go. Both of those things are
- 12:22critical factors in implementation.
- 12:24When you say see what the system is doing, can you elaborate on
- 12:27that? Yes, one of the most critical
- 12:29things with any communication system is is you can run
- 12:32analytics and and you can start to see things like word cloud of
- 12:35was someone angry, right? Did someone, you know, say
- 12:38something that they shouldn't have, right?
- 12:39Are they happy? And so digging into that and
- 12:41seeing what those interactions actually consisted of and
- 12:44listening to the call if you need to.
- 12:46So that gives you the ability to run quality metrics against your
- 12:50environment, which usually is reserved for very large
- 12:53corporations. That is one of the biggest
- 12:54barriers to entry to having anybody, whether it's an agent
- 12:58or an AI interface, in this case answer calls, is understand that
- 13:02quality behind the scenes. And because of the AI, I'm
- 13:05assuming that this turn around is very fast.
- 13:08Almost immediate the call is recorded if you want it to be
- 13:11and then the call is is brought forward.
- 13:13And then we, we actually use AI to do a summary analysis of that
- 13:18and it almost immediately you will see what happened and you
- 13:20can drill into that and diagnose what, what occurred with that
- 13:24interaction. Damon, what's involved with
- 13:26deploying this kind of system? The key here is to making the
- 13:29system easy to deploy and so making it practical to deploy.
- 13:32So what you would do is you would just start feeding in
- 13:34knowledge base articles. It can crawl your website
- 13:37information, information. You can even put simple keywords
- 13:39in there to say if you hear this, then say this, or if you
- 13:42hear this, transfer this to an operator or transfer this to a
- 13:45salesperson. So really it's it's very
- 13:48practical to set it up, it's very practical to have it learn
- 13:52with you and only the things that you feed into it will it
- 13:55actually be able to respond back to.
- 13:57What about integration with existing systems and existing
- 14:02business processes? A couple of things.
- 14:04First off, it's very important to be able to integrate with
- 14:06your human employees, right? We talked about the ability to
- 14:11capture the sentiment of a call and make sure that gets
- 14:13transferred onto a human. That's an important part about
- 14:16this. But the other thing is the
- 14:18ability to integrate with your systems.
- 14:19So if there is information we capture, right, be able to
- 14:22integrate that back. And that's one of the things you
- 14:24want to look for is how does it integrate with the system that
- 14:27you're mostly in? And usually that's ACRM.
- 14:29So making sure you've got the proper CRM integration, making
- 14:33sure posting that proper integration into the system is a
- 14:36critical component to make sure that you don't have to go back
- 14:38and forth between systems as you're as you're moving forward
- 14:41and working with your human employees.
- 14:44You've mentioned the data several times and the training
- 14:47of the system. Tell us a little bit more about
- 14:51what kinds of data are used. Simply put, any kind of data
- 14:56that is in a written format, in a digital written format, you,
- 14:59you can feed into the system. So again, you know, FA QS,
- 15:02knowledge base articles, those types of things.
- 15:05But the other thing is the, the notion of observability and the
- 15:07ability to, you know, go in and tweak and optimize the system.
- 15:11That's critical. Like to see what's happening
- 15:13with these interactions, see what's happening there.
- 15:15And then you can either tweak the documents accordingly.
- 15:18You put some additional prompts in there for that.
- 15:20That's something that's, you know, part of as an example of
- 15:22what we've done is we built that into the interface.
- 15:24You can just feed those documents in.
- 15:26You can, you know, you can tweak those those questions if you
- 15:30want to and all those kind of things.
- 15:32That's all part of the data that goes in to make sure you can get
- 15:34a quality response back out. Do you have recommendations or
- 15:38best practices for folks that are just starting out?
- 15:42Start with a simple practical issue you're trying to solve.
- 15:45So if you're getting the same call for the same thing over and
- 15:49over and over again, that's very tangible, it's very
- 15:52understandable. And you can usually put the
- 15:54information you want an AI receptionist to say back to the
- 15:58customer very, very quickly. So again, in the case of go to,
- 16:02we've allowed you to come simply come in there and put simple
- 16:05questions, simple answers. When you hear this, you can
- 16:08respond with that. And as you do that, what you'll
- 16:10find is you can tune that very quickly to the other things you
- 16:13want to do. You'll gain reliability, you'll
- 16:15gain trust, right? You'll gain the ability to get
- 16:18more confidence and then train that AI receptionist as you move
- 16:22forward. But starting with something very
- 16:23practical and something very simple is a great way to get
- 16:26started and and and makes it very real for businesses to get
- 16:30a handle around. Starting small enables that
- 16:34trust level to grow because it is new.
- 16:37I'll give you an example, Michael, of where we've seen
- 16:39something really work really well is so I have a medical shop
- 16:43that I work with and, and they have an after hours answering
- 16:45service. And after hours answering
- 16:47service is literally writing down each person that calls and
- 16:50then they're faxing that information in to the office the
- 16:53next day. Nothing wrong with that.
- 16:55That's happening today. But the air receptionist can do
- 16:57the same thing. It can actually take those
- 16:59calls. It can capture what they called
- 17:01about it, capture the person's name, depending on what you want
- 17:03it to do. And that's a very practical
- 17:05thing you can start with. That makes it easy for you to to
- 17:07work on and tune and adjust as you go forward.
- 17:11Can you talk about the collaboration between the
- 17:15virtual workforce and the people working at the company?
- 17:20It's absolutely critical to make sure that the AI workforce can
- 17:24communicate and can talk to and work with your human workforce.
- 17:28So you might want to, in some cases have the AI workforce at
- 17:32the beginning of a call. You might want to have them
- 17:34towards the end of a call, for example, if you wanted to ask
- 17:36for a post call survey or something of that nature.
- 17:38But making sure that you can transfer the calm intent, right,
- 17:42automate those responses that the that you want to as a
- 17:46business and make sure that we're not replacing the human
- 17:49because the human touch is still needed.
- 17:51And I don't see that going away anytime soon.
- 17:53But making sure you're augmenting what the human is
- 17:56doing by by taking those common repetitive tasks out of their,
- 18:00out of their view so they can go do something more special.
- 18:03That's the kind of interaction, the kind of speciality and
- 18:06specialist week that AI can bring to the market.
- 18:08You're really talking about redefining some of your
- 18:13processes to accommodate this new aspect, this new virtual.
- 18:20Team, yes, great way to look at it.
- 18:22As you take those mundane tasks out of your workflow, what can
- 18:28those employees be doing that provides better value to the
- 18:33company, that provides, you know, higher quality
- 18:35interactions with the customer? Because those things will always
- 18:38be there. People will always be needed.
- 18:40But taking some of those things that you know, in some cases can
- 18:42be frustrating because you're answering the same question over
- 18:45and over again. Automating that and then making
- 18:47sure you can work with the human.
- 18:48At the end of the day, those are critical aspects to making sure
- 18:51that we have that human to AI interaction nailed.
- 18:55Do you have any advice on on this notion of redefining the
- 19:00roles to work successfully with digital workers?
- 19:04What can you do to delegate administrative tasks as an
- 19:08example or things again that you don't want to do?
- 19:10So One really tangible example of this is if you take a call,
- 19:15at the end of the call, someone needs to summarize the notes
- 19:19into the CRM as an example. And what I can do for you very
- 19:22quickly is summarize those notes for you.
- 19:24Now you can of course go in and edit those if you want to, but
- 19:27having to be posted in for you automatically is a summary that
- 19:30you can accept or you can edit later.
- 19:32That is an example of an administrative task and an
- 19:34overhead that you probably don't want your employees to have to
- 19:37worry about. As they do that, then they can
- 19:39actually start to put their time and attention towards other
- 19:42things. So I think it's critical to look
- 19:44at those, at those parts of it and then figure out what the
- 19:46opportunities are that maybe you haven't been able to look at
- 19:49before with your employees that that they can now lean into.
- 19:52So again, you're really starting with simple labor saving use
- 19:57cases. Correct.
- 19:59Exactly. What about privacy and security
- 20:01concerns? First of all, make sure that
- 20:03you're using cloud systems and AI systems that have published
- 20:07security policies with, with, you know, encryption, fail
- 20:11safes, things of that nature. Can't say enough about that.
- 20:13And making sure that those things are happening.
- 20:15Usually that's happening with the commercial vendor, but you
- 20:18want to absolutely double check that that's that's the case.
- 20:20You also want to make sure that you tune the AI to respect your
- 20:24privacy policy. So if there's things you don't
- 20:27want the AI to ask, if for example, if your medical, you
- 20:30probably don't want to ask for certain patient information, you
- 20:33don't want to ask for things like Social Security number.
- 20:35So tuning your security policies with the AI to make sure that it
- 20:39doesn't ask for things that you don't want to get access to.
- 20:42Make sure that not only is the technology safe for you to use,
- 20:46but you're keeping your customers and your business safe
- 20:49from anything else that might come in and be captured
- 20:51accidentally as part of the system.
- 20:53So you protect the company and then also the users because
- 20:57there is a, there's a natural curiosity about how AI will
- 21:03handle our data and our privacy. And so you can secure that by
- 21:07making sure that your customers understand it and you're
- 21:09transparent with them about what you're capturing and what you're
- 21:12not through your AI systems, which is a critical component of
- 21:14success. Damon, where is all this headed?
- 21:17We will see a much, much better personalization and the ability
- 21:21to interact with people as if they were talking to a person.
- 21:26So imagine as an example, if you called and do like, let's say an
- 21:29automotive dealer as an example. I know you own a, you know, 2024
- 21:33Toyota 4 runner, right? I can actually bring that
- 21:35information in and, and ask and say, are you calling about that
- 21:39particular vehicle, right? Or I can call, I can ask about
- 21:41something that's very, very personal towards you.
- 21:44And so I think that's one thing you're going to see the latency
- 21:47will improve. So the ability for you and I to
- 21:49talk and an AI to talk exactly like we are at the exact same
- 21:53pace will be very, very critical.
- 21:55And then I think the other thing you're going to see is we talked
- 21:57a little bit about agentic AI and the ability to take actions
- 22:01on your behalf. And so as tools emerge, become
- 22:04more integrated with your different systems, being able to
- 22:07take more actions on your behalf and make that more seamless.
- 22:10And that's really where you'll see a force multiplication for
- 22:14businesses where we can take it can take common actions, not
- 22:17just answer common questions, but take common actions that
- 22:21might span different systems in your environment.
- 22:23Those are things that are happening.
- 22:24Many of those are there today. It's only going to get better as
- 22:27we move forward. Exciting times ahead.
- 22:29Absolutely. Making it practical for all
- 22:32customers to be able to adopt is what our missions all about.
- 22:35Damon Covey from Go to Thanks for taking time to chat with us
- 22:39today. Thank you, Michael.
- 22:40Great to be here.