0:00:00 - (Alane): Welcome to Automate Your Agency. Every week we bring you expert insights, practical tips, and success stories that will help you streamline your business operations and boost your growth. Let's get started on your journey to more efficient and scalable operations. I think my go. One of the most popular AI agent discussions that we're having right now with customers is email triage support. And, and really it's encompassing. I'm seeing three things.
0:00:31 - (Micah): Okay.
0:00:32 - (Alane): Yeah, I'm seeing. So if we think of email triage, because everybody's like, oh my gosh, the amount of emails I get, holy cow. But email is one piece of customer support. There's. You might have a support ticketing system, you might be getting emails to help customers and answer their questions.
0:00:48 - (Micah): But there's also, hopefully you're getting emails.
0:00:50 - (Alane): You'Ve got some kind of interactions happening.
0:00:53 - (Micah): Yeah, let's. Let's hope your clients are actually communicating with you.
0:00:59 - (Alane): Yes. And then the last one would be voicemail callers calling in. And we haven't really seen. And I've been having discussions with some of my business owner friends that also create AI agents and really answering the phone and having the AI agent answer the phone and doing a voice conversation. We just are not seeing that. It's super strong at that yet. Will it get there? Oh, I have full confidence that it will get there, but right now it's just the technology's not there.
0:01:25 - (Alane): But if we think of a caller leaving a voicemail and being able to use that for triage support, I still think that that factors into those three ways that customer support triage can work.
0:01:37 - (Micah): Yeah. So maybe we do some definitions here. So triage essentially is just saying you've got a bunch of things coming in at you and you've got to categorize it, organize it, do stuff with it, handle it, process it. Any of those words are based. That's what we mean when we say triage. So you might have a bunch of new leads coming in via email from a form. You might have clients giving you feedback on projects that you've sent their way to review to your point, you might have incoming phone calls or voicemails left.
0:02:10 - (Micah): And when you're thinking the phone call side, you're thinking more. Hey, we already have the technology to get the transcripts from all of this, so that can be filtered right in as a channel and work perfectly.
0:02:25 - (Alane): And most software that you're using now for your, your phone system transcribes. So having, you know, if you're, if you're looking at one, make sure it does that and then make sure it has an API.
0:02:36 - (Micah): Yes, always, always an API. We need to like wear T shirts for our podcast, Alane, where it's like always have an API.
0:02:46 - (Alane): Yes, that's work with software that has an just is a really great rule of thumb.
0:02:52 - (Micah): Yes, yes.
0:02:54 - (Alane): Every startup now hates us.
0:02:57 - (Micah): Exactly. Before AI agents the only way to do triage was human. It was you're sitting there, you're paying somebody to sit there and go through it. Even in our last business, Alane, that was I know at least one person's full time job and I think we could probably identify some part time responsibilities across the different departments fielding work. Now with our last business, Alane, we had at least one full time position was 100% triage. Right.
0:03:31 - (Micah): Tickets coming in. Well what team do they go to? Because we had multiple teams for all the different clients that we were managing. So routing it, resourcing, following up all of that stuff. Humans but with agents, agents excel at this. So if you have your phone calls coming in and it's being transcribed, that's perfect for an AI. If you have an email, if you have a support ticket, even at any other channels that can all come in and now get automatically triaged.
0:04:04 - (Alane): And I, I see there's two ways that the, the triage can work and, and can work together and maybe there's others maika that you're thinking of but there's one just very surface value is having it go to the right person or team. So it's filtering where that should go immediately and then the other side is already pulling the resource to answer the question. Now a hundred percent, you know, a human needs to be involved. I think still at this point is you know, making sure that it's answering it correctly. But you're saving. I was just talking to somebody, I can't remember who it was and their daughter just got a new job and she was handling, I can't remember exactly but it was like a hundred tickets a day.
0:04:47 - (Alane): Now she can handle a thousand a day because instead of manually writing out each response, she's reviewing the responses. And so super quick, you know, it filters to the right person that's responsible with a resource that can already whether it's a drafted email response, drafted ticket response, where it's pulling from a knowledge base of articles or something that you already have internally and can assist your team with that.
0:05:12 - (Micah): I mean I think this is something just pinpointing on this thread a little bit Alane, setting the expectation of what AI should be doing because it, it could be responding should it be responding? And I think that's. That's a pretty interesting thing. I think as we started developing stuff like, I, I definitely live in the bigger picture world of, oh, man, we could do this and we could do this and we could do this. And there's strengths and weaknesses to that type of thinking.
0:05:43 - (Micah): And I've had to kind of dial myself back a little bit on this because, yes, AI could reply to all of these things, write the messages, do all this stuff, should it? And I think dialing expectations back to say, wait a second, we are gaining so much value, we're gaining so much efficiency, we're gaining more capacity by just having AI categorize and assigned to the right person. If we just stop there.
0:06:11 - (Micah): The cost savings and the return on investment for implementing a system like that so that your team members can do things quicker and better, it's huge.
0:06:23 - (Alane): It. It is. Because if you're thinking about how your system works right now, it. It is bottlenecked by a person. Right? They can't do everything at the same time. So there's some type of. I'm going to call it a backlog. Whether their backlog is getting through their emails, they might be on meetings, they might be on other calls, they might be helping other customers. So there is a backlog of support that gets generated.
0:06:46 - (Alane): Whether you have a team of people that are handling that or it's all going through one person, there is a bottleneck happening that doesn't need to be there any longer. If a customer, what happens if they're sick? Well, yeah, and then you have to fill in that spot. But the value is in getting the response to the customer. It isn't how it gets there. So the faster it gets to the right person, that is the value you are providing.
0:07:12 - (Alane): So if the person is calling and it's a billing question, then that can automatically go to the right person. We can identify. This is a billing question. Great. This needs to go to so and so or this team that has access to this inbox, whatever it might be. That can happen immediately.
0:07:30 - (Micah): Yes, yes. It's that alone. If, if that's all you did right now, do it. Because that, I mean, we even use this internally. I think we've talked about this in other episodes, but we have an AI agent watching our email for errors and warnings and maintenance updates. Because previously we had to have somebody do that and then go, okay, which client is this? Okay, you know, how do we resolve this? Okay, when is this downtime happening? Okay, do we need to reset this automation because the Eric whatever it is, right.
0:08:04 - (Micah): That is 100% triaged right now. An agent sits there, watches for those to come in, finds the client in our project management system, sets up the task, leaves the comment at mentions the right person, assigns a due date, sets a priority, all of those things done.
0:08:21 - (Alane): So I want to. You said that so fast, Micah, that I get excited. I know, because it is so cool. I mean, I was just telling my. I'm in EO Nashville. I was just telling my form like, we are building the coolest things and it's so exciting to see. But what, what you just said, no one is watching for those errors except for an AI agent. It is identifying the customer that the error came in for, matching that to our ClickUp account for that customer or not the account, but the project for that client, and then pulls in the right team members that handle that client.
0:08:55 - (Alane): Hey, this is happening. And if multiple errors come in, the AI agent starts to put some urgency. Hey, yes. And it's wild. Like, hey, this has happened four times. Like, we need to jump on this crew, get in here and do this. And it's not necessarily anything with. With us, but that the make error. Something is not happening through the APIs. The API might have gone down for one of the sites. So, hey, when this gets working again, we need to be there to run these scenarios.
0:09:24 - (Alane): Our team didn't have to do that matching. If it takes 20 minutes a day for each one of these, for whatever your team does think about that is 20 minutes. If you distill everything down to 20 minutes. Because at a minimum, that's what it takes. Because their attention just got pulled away. They had to go and take that information, which is data. They had to take that information and put that data in another place, pull in.
0:09:47 - (Alane): So at a minimum, it took them 20 minutes because then now. Okay, that's done. And now I need time to get back into the work that I was doing.
0:09:53 - (Micah): Get back to whatever they're doing.
0:09:54 - (Alane): Yeah.
0:09:55 - (Micah): Or I mean, even. Right, right. In addition to this, Alane, the whole concept of, well, what if the person who's normally watching this is at lunch and an error comes in and it's urgent that isn't getting addressed for an hour because they're eating. Which people should. We should have people go to lunch. Yeah. There should be breaks. And so. Yeah. Or in meetings. If you're in meetings for too long. Oh, man.
0:10:22 - (Micah): We didn't actually get to this urgent error that was coming in now. Don't worry about that at all.
0:10:28 - (Alane): I mean, that's what I think about a lot of times the wait isn't what the time that the customer is waiting isn't the time to solve the problem, it's the time to get it to the person to solve it.
0:10:39 - (Micah): Yes.
0:10:40 - (Alane): And so most times there's. There is a one to one relationship. Most times, whether it's the account manager, the receptionist, somebody they go to for help, that account manager typically does do a lot of things, does answer the question. But there are times where they aren't the right person. Maybe it is a billing question that it could go to finance to help and they could get their answer much faster.
0:11:04 - (Micah): Or a technical question to the development team or a product team.
0:11:07 - (Alane): Yeah. Like I think of one. I think about this all the time because it was such a big question. We got. And we had an email template for this is. Where are all my Yelp reviews? I got. I had a hundred yesterday, now I'm back down to five. Well, it's because they all got filtered. Okay, well what's the filter? Where do they go? And so we had this. But that's something. If 15 years ago, if we would have had AI agents available, it would have identified. That was the question. It would have pulled the email template and already been able to respond to that client. Hey, this is why the AI agent would have had access to that email template which explained everything, even with a video in there and sent it on its way.
0:11:48 - (Micah): And so what you're talking about now is a transactional thing. And when we talked earlier on the episode of should an AI respond? Yes, this is one of those cases. Yeah, this is one of those cases where it's probably fine now. You need experts to help you build this and test it. Don't just go and try to build this and hope that it works. Right. There's definitely resources and stuff for that that we're putting together.
0:12:16 - (Micah): So we've got a mastermind, we've got a number of courses that we're putting together to help with all of this. You can always reach out to us with questions on these things. But I say all of that not to just shamelessly promote ourselves while you're listening to this, but also because please don't try to have AI automatically do stuff that's client facing or customer facing. Without the right testing, you can get yourself in a world of hurt.
0:12:41 - (Alane): Yeah. And even if you leave the automatic send out that email triage could identify that support ticket, have access to your knowledge base, know how to respond, and it can sit there as a draft for the human to review, tweak it and send it on its way. But that took off. That took care of 95% of the work.
0:13:00 - (Micah): Yeah. And if you really start thinking about this from a category perspective, what are the channels that are coming in? What are the types of things that you get? So what you talked about, Alane, was something that we got over and over and over again. And we built a manual system to a degree to handle that with our last business because we didn't have the tools that we do now, but now we do have those tools. So we could say, hey, when an email comes in and it's this type of email, reply to it this way.
0:13:32 - (Micah): Draft a message and we can even have it do things like, all right, what team would normally reply to this Customer success? The red pod. Fine, let's auto assign a task in asana or ClickUp or Monday that, hey, please review the draft and send to the client. Boom. It's not like we're talking about maybe a penny's worth of expense. And five seconds like this, it's, it's mind boggling compared to the 20 minutes of, okay, now I've got to check this email. Okay, now I've got to find the template. Where was that? Now I've got to edit it. Now I've got to reread it to make sure that I didn't typo anything. Now I've got to do this. And that might even be prior to somebody triaging it, then getting it to the right person, then getting the template. Like if you think about your workflows and all the steps that are happening just coming in from internal requests or external requests, new orders, new leads.
0:14:28 - (Micah): There's so many steps that can be streamlined with this technique that we're talking about right now. The email phone. I don't have a good name for this. We'll just call it the AI. The Triage agent.
0:14:41 - (Alane): Yeah, the Customer support Triage agent. We'll give it the longest name possible and then possible.
0:14:48 - (Micah): Yeah, well, I was going down that path for sure.
0:14:51 - (Alane): All right, anything else to add, Micah? I think even if a company just like you said focused on this as the first AI agent to build it would not only make you more efficient, but the cost savings that you would be able to get out of it.
0:15:06 - (Micah): And I mean, yes, let's put that in perspective real quick, Alane, because I don't think there's any way that you could challenge the cost savings on this. Even if it takes a person five minutes per email, like whatever you're paying that person is even for five minutes is going to be more than a penny, and five minutes is a lot more than five seconds. And then you look at the quantity of emails, the quantity of phone calls, the quantity of inquiries, the quantity of forms, the quantity of internal requests. Like, if there's one thing you do for AI agents, implement this first.
0:15:42 - (Micah): And it is the starting point to go. Okay, now that we have it categorizing stuff and assigning to the right person. Well, should we have it create a task? Should we have it draft an invoice? You can even have it review attachments and determine what is this attachment? It's. It's crazy what this can do and the value that this can provide to get people the information they need to do their job. That's the key to all of this.
0:16:09 - (Alane): Yeah. So we can be creative and strategic.
0:16:11 - (Micah): Yes. So starting point for this, I would say, is start thinking about your channels and start thinking about the categories, the types of inquiries, the types of emails, just the types of requests. Start categorizing that. How does that fit together? And then what is the routing for that? If you have that, you have 80% of what you need to be able to get something like this built for your business.
0:16:40 - (Alane): Yeah. And it can be relatively quick at that point because you got all the buttons ready to button.
0:16:47 - (Micah): Yes. That's how it works.
0:16:49 - (Alane): You can button up.
0:16:50 - (Micah): Yeah.
0:16:53 - (Alane): All right, perfect. So if. If you want to talk more about this, we'd love to chat with you. Like Micah mentioned, we're launching a Mastermind so that we can lead conversations with you on these topics and you can get ahead of the game. Notify us if you want to get added to the wait list. And thanks again. Thanks for listening to this episode of Automate Your Agency. We hope you're inspired to take your business to the next level. Don't forget to subscribe on your favorite podcast platform and leave us a review.
0:17:19 - (Alane): Your feedback helps us improve and reach more listeners. If you're looking for more resources, visit our website at biggestgoal.ai for free content and tools for all automating your business. Join us next week as we dive into more ways to automate and scale your business. Bye for now.
