Why AI Automation Fails, and What to Fix Before You Automate Anything

Ty Givens: The most expensive mistake that we all make in those types of spaces - and I'm including myself - is time. We make the mistake of burning up too much time thinking that we're going to save time. My recommendation is take a beat, think about what you want to do, and build from there.

Or I'll say this: treat yourself like you would a client. I would never approach a client's need in the middle. I'm always going to sit down and strategise for them. So I would say sit down and strategise for yourself.

[00:42] Mike Allton: Greetings program. Welcome to The AI Hat Podcast. I'm your host, Mike Allton.

I'm absolutely thrilled to introduce Ty Bryant-Givens, founder of CX Collective, and the operational mastermind who's been called into companies like Thrive Market, ShoeDazzle, and See's Candies when their systems are completely broken and their teams are drowning.

Ty has this incredible superpower. She can walk into operational chaos and transform it into a well-oiled machine that actually works for humans before any AI touches it. Her philosophy - AI doesn't solve chaos, it scales it - has saved countless businesses from throwing good money after bad tech, and today she's bringing that same wisdom to help you build the foundation that makes automation actually work instead of amplifying your headaches.

Hey, Ty. Welcome to the show.

Ty Givens: Mike, first of all, thank you. And let me say, I am going to use your intro henceforth for anything else that I do. Oh my gosh, that was awesome. Thank you.

Mike Allton: My pleasure. And I'll always be transparent, Claude writes my introductions - but to your point, Claude has been trained by me to know that I don't want to just read somebody's bio. Even if they give me the bio, which most guests do, I don't want to just read it. That's boring. I want to hype them up. I want to transfer energy and excitement to them on behalf of the audience.

Ty Givens: I call it a win, 'cause I'm hyped now. So, yay. Thank you.

[02:12] Mike Allton: Awesome. Well, I know you've seen this story a hundred times. A business throws money at AI tools hoping to escape operational chaos. What do you think's the most expensive mistake you've witnessed when someone tries to automate before they organise?

Ty Givens: Listen, I'm just going to shoot from the hip. I do it myself. I have been out on my own for almost ten years, and sometimes I'm doing things and I'm like, "Oh, you know what? It would be cool if I had a process around this, if I automated it." So I use Make.com to do a lot of my AI agents, and I might go in and start building that agent without a proper foundation. And what ends up happening is I spend so much more time fixing and tweaking and editing and changing than I would have if I had just sat down and said, "What do I want it to do?"

And so I would say that the most expensive mistake that we all make in those types of spaces - and I'm including myself - is time. We make the mistake of burning up too much time thinking that we're going to save time. My recommendation is take a beat, think about what you want to do, and build from there.

Or I'll say this: treat yourself like you would a client. I would never approach a client's need in the middle. I'm always going to sit down and strategise for them. So I would say sit down and strategise for yourself.

Mike Allton: Love that. And I fell in the same trap just a month or two ago. I wanted to explore Zapier's AI agents. Two hours later I had built nothing, and just increased my frustration level and wasted some time. There was a valuable lesson packed in there, I was advised - but still, it felt like a bit of wasted time, because I didn't come at it with a specific business problem or process that I was trying to solve or replicate. I was just playing around.

**[03:37]** So walk us through your philosophy - AI doesn't solve chaos, it scales it. How do you help someone recognise that they're about to make this costly error?

Ty Givens: One example is to realise that if you don't go into it with a plan, what you're going to get out of it is likely just going to be hecticness. That's not a word. It's going to be chaos. That's the reality of the situation.

I know that sometimes we don't know that we're in that deep until we get in it. I too have a couple of Zapier AI agents that are just sitting, waiting for me to do something with them, and I have not yet, because I have to be in the right mindset.

So what I would say to that is, you need to recognise that if thirty minutes in you can't see the light at the end of the tunnel, it's probably time to just back out of that tunnel and sit down and make your list. I'm a big fan of making diagrams, just so that I can understand what will happen in more than one instance.

'Cause I find that when I make a list, it's linear. This will happen after this, after this, after this. But the reality is, when you're building out any resources or any agents or tools using AI, you have to have more than just one outcome as a possibility. So what happens if this happens? Okay, and then if this doesn't go as planned and that happens, then what do I do? And so forth and so on.

So I would say build a chart, a diagram. Help yourself to really think through different scenarios and situations, so that you don't end up in that burial ground of a dozen half-started AI agents that are doing nothing, and a lot of wasted time that you'll never get back.

[06:03] Mike Allton: Yeah, that makes a lot of sense. I know one of the things we're talking about here - we're talking about AI agents, AI assistants. A lot of people, myself and I think yourself included, talk about treating them like employees. What do you think that actually looks like in practice, particularly business owners? How do they have an actual onboarding process for AI tools?

Ty Givens: Yeah. I literally treat it the same way that I would my human assistant. I provide context, I give examples, I ask questions, and I course correct along the way. I'll tell it when it's wrong.

**[07:07]** And another thing that I'll say is that I've learned how well I communicate because of how the AI products respond to me. So I think that there's a two-way street on that. You're going to learn a lot about yourself and how you move as you are trying to educate this product, or this tool, on you.

It can also be difficult if you're someone who is not so good about talking about yourself, or is not so good about setting expectations. And I say that because I have this theory - and I'm happy to be wrong, anyone can prove me wrong, I'm sure - but people who are super good at social media, that is their job. People who are really good at their job are good at the job, and probably not as good at social media. So you're going to find that when you're trying to use this tool, it's going to expose what you yourself are good at, or what opportunities you also have in how you deliver information to other people.

I've even used AI to critique my communication style, and to critique how I interact and engage and approach projects. And it was quite humbling. But it was also something that I used to learn about how do I want to show up, and what changes do I need to make to do that. And it had actually helped me to do that too.

While I talk to it like it's human, I'm not one of those people who feels like it's my friend or anything like that. What I recognise is that it gets the rawest, most sincere and honest version of me and what's in my brain, and I use that as leverage from which to build and improve.

[08:49] Mike Allton: Yeah, that makes a lot of sense. I've reflected several times about how I feel like I'm very good at using AI, and I feel like that stems from the fact that I've been a well-trained, very good communicator for several decades now. And I was also a computer programmer. I was trained on how to identify and solve problems. And I think that happens to weigh in a lot here.

So speaking of solving problems - I know you work with a lot of other companies. When you're first walking into a new client, how are you going about auditing their processes, auditing their problems? What's the first step that you take when you're walking into what's probably a very chaotic operation? And the second part of that is, how do the listeners today apply that?

Ty Givens: Sure. So the first thing I do - we call it the investigate phase. One thing I have to do immediately is shake off any assumptions or preconceived notions about what they're doing right or wrong.

And so I would say to the listeners, do the same for yourself. Give yourself that grace to say, "This is just how I'm operating right now. I can always improve." It doesn't matter how good you are, there's always a way to do things better. So just stay in that mindset.

But I want to understand from the customer's perspective. So - it's me and my team - we look at things from the outside in. We start with something as simple as the help centre, if they have one, or FAQs that they have on the website, to see what type of information they're communicating to the customer, how that information shows up. Is it useful? Is it fluffy? So we start right there.

And then the next thing we want to know is, what are the policies? What have you told the customer are your policies or guidelines? And then on the inside, what have you told your agents that they can do above that?

The reason why that helps, especially from an automation standpoint, is the AI tools and products - as many people probably already know - are built based on content that you develop. And most people don't spend a lot of time developing the content. And we actually start right there. What are you telling your customer? What are you telling yourself?

Once we identify what we're telling the customer and what they're telling themselves, then we start to match that with the vision that they have. And from there comes a set of intentions that we put together to help the company start to move in the direction that they want.

Sometimes they will decide to have us do a full implementation. That actually happens most of the time. But there are instances where a client will just say, "Tell us what to do, and we want to take it on, and we'll let you know if we need help after."

[11:13] Mike Allton: I love how you started that by framing the entire investigation as - this is not judgement. This is not right or wrong. This is just, this is what you're doing, or what we're doing, or what I'm doing right now. And I need to just think objectively: does this need to change?

I've found that to be a very freeing perspective as I'm working. I have an AI chief of staff. It's just me - the dogs don't help me at all. And it knows everything that I've told it it could possibly know about me, my business, my goals, objectives, challenges and limitations, and so on. And it's very freeing and liberating to me to have a conversation with it, strategic about the business.

I almost try to disconnect myself from what we're doing right now, whether it's who we're targeting, the campaigns that I'm running. Say, look, this is what I'm doing. This is what we're talking about, or who we're talking to. Let's pull this apart. Let's unpack this. Let's see what's working.

**[12:26]** But from your perspective - like I said at the outset, you've worked with some really major brands. Thrive Market, See's Candies. What do you think's an operational mistake that they're making that businesses can learn from?

Ty Givens: So let me first clarify. Thrive Market, See's Candies, Intuit, Herbalife, Office Depot, Home Depot, all the Depots - I was an employee there. And a lot of what those companies gave me was a foundation to know and learn and understand what it actually means to operate at scale.

Thrive Market was a startup, so I was the seventh person brought into the support team - but I was brought in as the head of that team, to grow and build that team. One of the biggest lessons that stuck with me from Thrive Market is something I talk about a lot, which is to be comfortable pushing back and saying no.

And what I mean by that is, the direction that I was given was, "Hey, we need to hire thirty people." But what I should have said is, "Yeah, I understand we need to hire thirty people. But I, as one human, cannot manage thirty people, plus be in the past, present and future, and work across people, process and technology. That is impossible. But what I can do is bring in leadership to help me build this team out." And I didn't say that.

So I would say, from the perspective of startups - especially a fast-growing startup, where a lot of people think, "Oh, let me join this startup, they're growing really fast, it's going to be a lot of opportunity" - the reality about fast-growing startups is that what they need is expertise. It's not really a space that you can go into and learn as you go. It's difficult to build a plane that's flying that high and fast. If you want to learn how to do things, or get opportunities to work on things you would not have, you want to go to a startup that's steady, that has a little bit of wiggle room. Thrive Market wasn't the place for that.

With See's Candies - See's Candies at the time, we called it a ninety-year-old startup. The website was new. They were just trying to modernise. The operating system that we were using was from the '70s. I don't know what they're using now, but things were so antiquated back then. The POS system was from the early 1990s. Everything was like going into a time warp.

And so I would say that their biggest opportunity was modernisation. But at that point, you're not turning a speedboat, you're turning a cruise ship. And you're not just turning the cruise ship in the direction of innovation and technology - you are also turning people who've been with that company for twenty, thirty, forty years along with it. And that is much harder to do.

**[15:16]** For other companies who I've worked in as a consultant - I've worked inside of over fifty companies at this point in my career - one of the most universal challenges that I see is an issue with process. And for the company, they don't realise that it's an issue with process. They think that their systems aren't working. Meaning, "Oh, we're using this particular help desk." A help desk example would be a Zendesk, or - depends on how big you are - you might be on Gorgias if you have an e-commerce or Shopify.

What they'll see is that this system is not working. But when we get in, we do that investigation, and we take the lid off, what we realise is it's not actually the system that's not working. It's what you told the system to do that's not working.

It's thinking about all of those things that you need to set up and organise. How do I want to tag tickets that come in? How do I want to categorise tickets that come in? What do I want to automate? What can I tell people that's always going to be the same thing? These are all the things that you have to think about, whether you're programming a system that people are going to work in, or AI agents are going to work in. And I think that pre-work is really what gets people boggled - because somehow they think, "I bought this new thing," or, "I'm paying for this new thing, it's going to solve all my issues," when in reality it's just another way to get something done that you now have to master.

[16:22] Mike Allton: So Ty, let's drill down a little bit into what you were just talking about. You're talking about companies that have FAQs, and they've got documented processes, and they're using systems like Gorgias or Zendesk and so on, which are all great. That seems to be where a lot of professionals kind of fall apart.

How do you create systems that capture institutional knowledge when you are the institution?

Ty Givens: I'm still working on that. No, I'm just kidding. There are plenty of tools out there. One of the tools that we're using to bring everything together - and this is not an endorsement, I'm just sharing my tech stack with you all - we are using Guru. And the reason why I selected Guru is because we use a product called Teamwork for our project management, and we obviously use Google Drive, and we use Slack. So Slack is how we communicate with each other.

And then we use a tool called Front. I know, you're like, "Oh my gosh." But we use a tool called Front, which is a help desk - but I really love it for B2B, because it looks like just a regular email system. It does give us the ability to share messages, to comment on messages. It has a lot of functionality. And their head of support there is amazing. He and I engage on LinkedIn a lot, so I feel very supported by them.

And so with Guru, we can bring all of those resources that have information - various pieces of information - into one place. And using a chatbot, we can engage with different pieces of information in one space.

So for example, every Thursday - today happens to be Thursday - I reflect on the week. What did we do? And then I think about the next week. What do we need to do? Before, I would have to go to different places to remind myself: what did we work on, what did we do? And check in with different team members to see where we are with things. And now I can just ask Guru, "Hey, summarise the week for me, along with the assignees, and let me know what's still pending and what was completed."

And that has saved me a good forty-five minutes per client of work of summarising information, because it's all right there. Now, we all know that AI isn't perfect, so I do have to validate it. But validating it is far easier if I just say, "Include links to the tasks," 'cause then I can just go in there and look at it myself and make sure that it's right. And I can also correct it.

The way that I look at the cost of any products or tools that I use is simply as an ROI. Whatever it is, it could cost me forty-nine dollars a month per user or twenty-nine dollars a month per user - but the amount of time it saves me, and also my clients, because they're not coming to us to bill a bunch of hours doing admin and research work, they're coming to us for expertise. So the fewer hours I can use researching, finding information for them, the more valuable we are to them. And the easier that I can make it for my team to find information, the less I bottleneck. So that's what we do.

[19:27] Mike Allton: Yeah. No, I couldn't agree more. But let's shift gears a little bit and talk about processes, 'cause that also, I think, is an area where a lot of professionals are lacking. They have their things in their mind that they normally do in different scenarios, but they're maybe not documented, they certainly aren't optimised.

So what are some non-negotiable processes that you think need to be just absolutely bulletproof before someone even thinks about automating, say, customer support or operations?

Ty Givens: Yeah. So depending on the type of business that you're in, you need to have clarity around when you would offer or issue a refund - money back, if that's what you'll do. Some clients decide that they're not going to issue any money back or any type of refund. Instead, what they will do is give you credit, or give you another month, or whatever that is. So having that be something that's documented, clear and solid is important.

And speaking from experience, because I oftentimes will have a soft spot for a client. This recently - I had a client who said, "Okay, I think we're okay." We were supposed to renew her contract on the sixteenth. She goes, "Is there any way we can cancel this?" This was about a week ago. Normally I require two weeks' notice, so that I can wrap everything up and get everything good to go. She was down to one hour of time. I was like, "Yeah, don't worry about it. We'll just go ahead and close it out."

Ended up costing me my own personal time, because I operate with integrity and so I don't want to tell you, "Yes, I'll do this," and then now I'm going to send you a bill. I just don't move that way. So anyway, I ended up having to own that. And if I would've just followed my own policy, that would've helped - which, AI agents, things like that, would be nice to lean on. I should have done that instead of taking it on myself, 'cause I have emotions and the agents don't. They just follow your process.

So yeah - knowing when you're going to end a contract, stop a contract, when you will issue a refund and what that's going to be. That's one.

What is your replacement policy? What are you going to do when a product is not delivered? How many times does a product have to be delivered inappropriately or incorrectly before you will say, "I'm no longer delivering to that address"? Little details like that.

How do you handle negative social media engagements? Most times people reach out to support, they don't get what they want, they go to social media, they want to get loud. How do you handle those sorts of situations? These are things that you need to think about - kind of like a prenuptial. You want to think about these things before you get into it. 'Cause then the emotion takes over, and now you're being mean instead of doing what's right. So you want to get ahead of those little details that seem really minute but end up being a really big deal later.

Shipping policy - how long will you wait before you drop an order to be packed? That is a big one. We work with a lot of e-commerce brands, and you'd be surprised at how many e-commerce brands will allow you to place an order but not be able to edit or change that order. Even if you call right away, most of the time they can't - it's already dropped into the WMS. They have their own process, they can't stop it. So making decisions like, "Hey, I'm going to put an hour buffer there," or, "I'm going to put two hours," or, "We're only going to pack orders from the previous day, the next morning." All of those little things that you just kind of think just happen - if it's your company and it's your business, you have to decide all of that stuff. And that is not easy.

[23:12] Mike Allton: No, it's not easy, but to your point, it needs to be done. These situations are going to come up. I obviously laughed and loved your example about social media, 'cause you're right. Those things happen. People complain on social media. Sometimes people are unreasonable - maybe they're not even a customer, and they're just giving you slack on social media because they want to.

Having a process in place does a couple of things. First, to your point, it allows you to not get emotionally involved in that conversation, in that moment, and to just follow your process. "This is how I'm going to respond to this scenario."

But to help us connect the dots - if I'm reading between the lines of what you're saying, it also allows us to inform our future automations, or AI agents, or documentation. If we've got documentation on how we handle interactions on social media, and then something comes up, we could then go to the AI, who has access to the documentation, feed it the comment or the post or the direct message, and ask it, "How should we respond?" And it will give us a response based on the documentation. Is that what you're saying?

Ty Givens: I am saying that. And then - I use LinkedIn a lot. I'm a B2B firm, so it's to my benefit to use that audience, because that audience is my core customer, or client I should say.

And sometimes I might feel very passionately about a post that's made - and I could feel passionately, and I could feel completely opposite. And what I love is that I'll still write out those thoughts just how I want to say it, and then I can drop it into ChatGPT or some other tool and go, "Can you make this sound kinder? Can you make this show appreciation for their perspective but still be firm on mine?"

And that even happens when you get a message from a client that is not ideal, or they say something that you may take offence to. I feel like you can always leverage these tools to help you really soften your approach.

I don't consider myself the most polished, but I'm authentic, I'm real, I'm relatable. I am who I am. And so there are instances though where I have to just take a beat and process that, because my brand matters a lot. How people receive me matters a lot. And thank goodness for these AI tools that help me to present myself in the best possible light, even if what I'm saying is contrary to what the author may be feeling or thinking.

[25:39] Mike Allton: No, I love that. I couldn't agree more. So for someone listening who maybe they're recognising themselves in some of this chaos that you and I are talking about today - what do you think is the one thing that they should stop doing immediately? And what's the one thing they should start doing today?

Ty Givens: Okay. So immediately, stop blaming yourself. Nobody's perfect. We all make mistakes. Think about the fact that as you're building your business, your company, whatever it is that you're doing - if you want to stay solo, if you want to grow it - any decision that you've made, you've only made it from the space of wanting to do and be better. So first of all, forgive yourself. Any mistake you made, it was a mistake. You learn from it, you move on. That's the main thing.

The next thing that you need to do is take time to think. It's a small thing. Change your environment. It could be going to a coffee shop. Sometimes I go to Barnes & Noble and work from Barnes & Noble, because the environment is different. Sit outside, go for a walk - but change your environment. When you change your environment, you change your thinking, and that is going to allow you to be more creative.

I'm right now physically in my office, and because I'm here, my mindset is very tactical and focused on the day-to-day operations that I need to do in support of my clients and my team. But it could be as small as me taking my laptop to the living room instead, and suddenly I'm now freed up to have these interesting and creative ideas of how I want to move and operate better.

So what I would say is, allow yourself to feel whatever you feel, but just be like, "Hey, I did the best that I could given the resources that I have." That is what it is. On the other side, change your scenery so that you can change your ideas, and you can change your viewpoint, and you can grow your brand.

[27:20] Mike Allton: See, folks, this is why I wanted to have Ty on the show. And like I promised at the outset, it wasn't going to be just about AI, how to set up a bunch of AI tools. This was truly, I think, a mindset-changing conversation and ideas - particularly your last two points, because I could not agree more. Had honestly nothing to do with AI specifically, and just about how to be successful humans who are dealing with the chaos that we're surrounded by every single day.

So Ty, you've been absolutely amazing. Thank you. I think this has been such an important interview. For folks who want to learn more about you, maybe they want to connect, maybe they've got more questions for you - where should they go?

Ty Givens: The website is cxcollective.com. If you add /ty, there's a page there where you can learn more about me and even schedule a fifteen-minute intro. Doesn't have to be anything other than a hey. If you have a question or something like that, I'm happy to answer. And from there we can talk about whatever you want and go from there. But if you're just curious, do that - and follow me on LinkedIn. I try to post at least four days a week.

Mike Allton: Fantastic. Thank you, Ty. Thank all of you for listening. Of course, we'll have all of Ty's links in the show notes down below.

Till next time - welcome to the Grid.

[28:31] The AI Hat Podcast: That's a wrap on another episode of The AI Hat Podcast. I hope you've gained some valuable insights and actionable strategies to boost your business with AI. If you enjoyed this show, please subscribe, leave a review, and share it with your network. And don't forget to visit theaihat.com for more resources and to connect with me. Until next time, keep innovating - and remember, the future of business is human, and it's powered by AI.

---