Automate Your Agency

What are AI Agents and what could they be automating in your business?

• Alane Boyd & Micah Johnson • Season 1 • Episode 41

Share episode

Send us a text

Running a business is already chaotic enough. The last thing you need is to be buried under emails, data entry, and mindless tasks that eat up your time. That’s where AI agents come in. 

AI agents aren’t just another buzzword, they’re like digital assistants that can handle repetitive tasks, make decisions based on data, and free you up to focus on growing your business. Unlike simple automation tools, AI agents can learn, adapt, and improve over time, making them a powerful tool for businesses that want to work smarter, not harder.

Think about all the time spent on answering emails, scheduling meetings, processing invoices, or tracking customer interactions. AI agents can handle these tasks, ensuring consistency, speed, and accuracy while freeing you up to focus on strategy and growth. But here’s the deal: AI is only as smart as the system you build around it. If your data is a mess, your AI will be too. 

We’ll be honest, this isn’t just a “plug in ChatGPT and let it run your company” kind of thing. We’re talking about real strategies, real mistakes businesses make, and how you can start using AI to your advantage without overcomplicating it. 

Here’s what we’re getting into:

🔹 What AI agents are and how they can automate tasks in your business 

🔹 Why having clean, structured data is the key to making AI work for you 

🔹 The ways businesses are automating tasks and cutting down on manual work 

🔹 The biggest mistakes owners make that cost businesses time and money 

🔹 Where AI is headed in the future and what it means for your business 

🔹 And so much more 

If you’re still on the fence about AI, it’s time to think again. The future is already here, and AI is leading the way. Don’t let your business fall behind, get ahead of the curve now. 

Enjoying the podcast? Drop us a 5-star review, it helps more people find the show and keeps us fired up to bring you new insights every week. Don’t forget to hit subscribe! 

Disclosure: Some of the links above are affiliate links. This means that at no additional cost to you, we may earn a commission if you click through and make a purchase. Thank you for supporting the podcast!

For more information, visit our website at biggestgoal.ai.

Want more valuable content and helpful tips?

Explore our Courses/Programs:

Enjoy our Free Tools:

Connect with Us:

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. Micah I'm so excited about today's episode where we're talking about AI agents, what they are and how to use them. And I know when we first started talking about this months ago, I was lost.

0:00:30 - (Alane): It was such a hard concept to understand. Now we are building the coolest things and just so excited about what's to come.

0:00:38 - (Micah): I mean, I have to admit it wasn't that long ago that even with a tech background I felt a little lost with AI agents too. I remember talking to a couple of people that I know that are doing some AI stuff and I'm like, it's just like making a call to ChatGPT, right? Like what's, what's the difference? And yeah, so in today's episode we are going to talk about that difference because it seems like there is quite a bit of confusion around what is just going to ChatGPT and you know, chatting what is creating an automation with AI elements and what's an actual AI agent?

0:01:22 - (Alane): You know, I think the easiest way to first grasp what it is without explaining what it is is when AI first came out and everybody was so excited about it automating everything in their business. And you know, one of the big thing things were even like chat bots or having a resource to be able to have your team just be able to go ask a question and then go through documents or past history and it give an answer.

0:01:52 - (Alane): And really I just wasn't there. You really had to invest a lot of money into building a model for that and having an output and most companies couldn't afford that and it was really challenging. And now it's like all those things that we wanted to do or businesses wanted to do with AI and automation is what I'm seeing an output of having an AI agent.

0:02:17 - (Micah): Yeah, I mean just last week to kind of give a sneak preview of what an agent can do, we built an agent that would you can chat with, but then based on what you're asking, it can actually go search like a SharePoint or a Google Drive folder, find the right document that you're asking about, search that document and extract the information that you need from that document to help save all the time of, well, which folder is it in, what document is it, what's the latest version, what's where in this hundred page Document is this.

0:02:52 - (Micah): And it just does it in a very short period of time.

0:02:56 - (Alane): Yeah. I think about policies, especially in global mobility industry, where they've got different policies for their transferees and you might have 100 different policies that had been written for people that are moving and you could ask, hey, what was the policy for this transferee? And it would go find it and then spit it out for you. And thinking about sales contracts, going back and asking what was in a contract without having to dig through and find it. Oh, my gosh, so many hours saved.

0:03:31 - (Micah): Yes. Yes. All right, so let's maybe start at the beginning. We first have just chatting with an AI.

0:03:39 - (Alane): Yeah. And then we took that. So chat, like going to ChatGPT or Claude or Copilot, whatever. And you're chatting with it. Very manual process. Right, Micah. Where you have a question, you go to it and ask it.

0:03:54 - (Micah): Yeah. So in this scenario, you. You could spend the time to find the document, you could download the document, you could go to ChatGPT or Quad or wherever, you could upload the document and then you could specifically ask about the document. It's going to save you a little bit of time, but you're doing a lot of legwork. You're doing a lot of manual work and it's just not that easy.

0:04:23 - (Alane): So then we graduated to AI and automated workflows, where it could take information and then do an output and then you wouldn't have to manually go to ChatGPT. But there's a piece of it still missing there.

0:04:38 - (Micah): Yeah, with, with automation, with AI components, definitely helps. It improves it. Right. Like you can streamlining, streamline getting in some of the, you know, documents or a very specific document. You can send that, but you're also missing like the chat aspect and you're missing, like the flexibility of the output. It's going to give you one output, and that's. With any automation with AI, you're. You're literally just having an AI step inside your automation, which is cool. Like it can. There are places for that. You could definitely say, hey, I want to send this to the AI and I want the AI to process it this way. And then I want it in this output format.

0:05:23 - (Micah): And you could get that. But it's. It's very fixed. Right. And it's only one time through the AI, you send. It's like an API call, You send it out, it processes it and you send it back. And that's it.

0:05:37 - (Alane): Yes. Still really cool Use cases. We have an episode on how to use it in the sales cycle. So if you're interested in hearing just how to use AI as a part of a fixed workflow, still really cool things. But the AI agent piece took what we're doing there and now added a layer on top of it where you don't have to have a fixed trigger. It could be a chat, it could go back in time, back into something and review it.

0:06:06 - (Micah): Yes. Yeah. So an AI agent at, @ its core is you don't have to give it the step by step instructions so it can start processing logic on itself. For example, if you send if. Well, we'll keep going on our example that we've been talking about, Alane. We can create an AI agent that says, hey, look at all the files in this folder, determine which one the user is looking for, find the information the user is looking for and summarize the information the user is looking for back to the user.

0:06:42 - (Micah): Those are literally the instructions that you would give it. Instead of writing like step one, list all the files, step two, open the f, you know, files. There's no way you could do that with a normal automation because the AI is using its own reasoning to say, well, is it this file or this file or this file? And then it can go into those files, read some of it and then go, wait, that's not it, let me check the next one.

0:07:08 - (Micah): Read some of it. Ah, this is what the user is looking for. Pull that information out, summarize it. And again, we're not saying summarize it in this specific format, in this specific way, in this specific output. We're just saying give the user a summary and it's up to the AI to figure out the best way to do all of that stuff. All of those steps that again, you could never logically write in a program or in a step by step automation flow.

0:07:40 - (Alane): Yeah, and I mean, I like, I like the name of AI agents because it's represents a person of what kind of work they would be doing. They would be the one that would go in and dig through an existing document or existing timesheets and give a summary and analysis. You know, things like that. Like it really is that concept of having someone, and in this case an AI agent going to do that work that no longer has to be done by a person.

0:08:08 - (Micah): Yeah, and I mean it's, it's interesting you bring that up because, I mean, just earlier this week, last week, we've been talking about this so much, Alane, and we've started implementing agents in our own workflows for ourselves. And one of the things that we Realized is not only is this helping out the team that already exists, but it's helping us fill in the gap to do things that we maybe always wanted to do. But it wasn't worth hiring somebody just to do that. And we certainly couldn't just keep stacking it onto our current team's responsibilities.

0:08:42 - (Micah): So you just mentioned timesheets and analysis of data. Man, have we changed our lives in the past five days by launching the AI agents for ourselves that literally every day, very simple instructions. Go get the data from today or from the beginning of this month to today and find out if any clients have more hours than they should. What are the outliers? And based on your findings, send us a slack.

0:09:16 - (Micah): That's literally the entire set of instructions for this agent. And the agent goes to toggle our time tracking software. It pulls out everything, it structures it hours by client and then looks down the list and goes any outliers, any trouble, anything we need to be aware of. And what it finds is it'll go and post a slack message to our project managers. What I find really cool about this is we're not defining what the message is.

0:09:49 - (Micah): So every, every day at midnight, this agent runs and does this for us. Now we have been spot checking over the years. Our project managers get in, but we're all human. It's hard to remember to maybe check that much of a detailed assessment and analysis every single day. So this fills in the gaps, right? We wake up and we'll get a message from the agent that said all, all clients are right on track. But here's your top two clients of, you know, hour wise to watch out for.

0:10:24 - (Micah): Brilliant.

0:10:25 - (Alane): It's, it is, it's so, it's so amazing. And then we've, we've taken it also because we want to know that on a daily or weekly basis. But as owners of the company or as project managers, they might need to be looking at this from a bigger picture too. How does this month's hours compare to the previous month? I also want to see it on a quarterly basis because I want to be able to plan. If we are consistently hitting a max amount of hours and our company is growing, then we need to be planning for hiring.

0:10:58 - (Alane): If we've hit that two quarters in a row, then it isn't just our team, as, you know, lots of tasks happening. No, we actually legitimately have the hours to show for it and can make business decisions. And the other thing that we're tying in with it as we use QuickBooks and looking at our revenue and how that compares to our hours, and that is again, like manual work that we were doing and then maybe sometimes pushed off if there were too many other things that needed to get taken care of.

0:11:27 - (Alane): And so it does fill in these holes that you want to be doing as a business, but that it just kind of hits the back burner because you're managing clients or you've got a million things on your to do list and things that you absolutely should be doing, but you just didn't have the time to do it before.

0:11:46 - (Micah): Yeah.

0:11:47 - (Alane): Now what about the things that we're doing that are on the marketing side? Because running your business and having the AI agents is super cool. Some of the HR or policy things that we were talking about or contract things, also really beneficial time saving. But I love what we are doing for some of our clients on the marketing side.

0:12:10 - (Micah): Yeah, for sure. I mean, one of the really cool things about agents is that they have a visual capability now. So if you provide it with designs or images or screenshots, it can look at that and then actually take action on what's in that image. So, for example, if you uploaded a. Or if you provided an agent with a mockup or a website design, you could ask them, how would I improve this? You could ask it to, hey, capture all the sections and save that as a site map. You could have another agent that then looks at the site map and goes, well, here's all the cool stuff I'm going to write about it.

0:12:49 - (Micah): And that's not even touching on the fact that we can have it search databases, search the web and figure out different ways to collect data. All accessible data, not private data, to be able to get better at the topics that it's writing about. And I mean, all of this stuff that we're talking about, Alane, it's interesting because it sounds like, oh, we could totally just replace our whole team. But the reality is that's still not the case because the last thing you want to do is have an AI agent, let's say, go through mockups, write all the content, get it on the website.

0:13:27 - (Micah): Cool. But at some point, a human needs to review that and go through it and say, is this right? Is this the right tone? Is this the way that we want to do it? You know, I've given a lot of presentations on this topic recently. It's this whole concept of we're not as humans, we're not just writing from scratch anymore, we're editing. And this is that perfect case. Like you, you still want the humans to do the Design.

0:13:54 - (Micah): But from the design to the site map, can we get an easier way? Yeah, probably. From the site map to the initial content, can we get an easier way? Hell yeah, we can.

0:14:07 - (Alane): Yeah.

0:14:07 - (Micah): And then from the content to getting it visible in a, you know, preview website. I mean, Alane, this is one of the things that we ran into a lot or continue to run into is we'll work on something and you know, you and I will trade drafts back and forth. But then seeing it in a Google Doc versus seeing it in a final presentation or final webpage is very different. So if we could shorten all of those drafting moments to say like, hey, here's what we're aiming for. And then here it is on the website before it goes live, then the humans come in and go, actually that looks like garbage. We need to change this, this and this. And we can change it, but it gives us all of those shortcut steps. So we're not writing content for three months.

0:14:56 - (Micah): We're actually working on getting a website launched in three months.

0:15:01 - (Alane): Yeah, I mean, whenever we talked about AI initially coming out, people were so worried about losing their jobs and it really helped make their jobs better. You know, where they could spend more time on either being strategic or creative. Maybe some of the tedious work could help. And then when you introduce AI agents, is that on steroids? That idea is your team doesn't have to do the little time consuming things that they could be better spent at a higher level doing, you know, other things in the business that provide more value. And the AI agent could do these little things for you.

0:15:39 - (Alane): And when you think about the companies that embrace this, you already started to see this gap over the last couple of years where the companies that are using AI are moving fast and making themselves more sustainable going forward and growing. But the companies that then now take AI agents and are using it are going to continue moving so rapidly compared to the companies that are not in their day. And being, you know, that's such a scaling right now and growth at a, in a lean way is so vital for companies.

0:16:16 - (Micah): Well, I mean, just think about the examples we already shared, Alane. One of the ones that we've instituted internally again for ourselves is the ability for the agent to watch specific emails coming into a specific inbox and what they're looking for are automation, errors, warnings, notifications of downtime, different things like that. Well, previously we had to have a human watch that inbox and then be able to figure out, look at it and go, oh, that's Client X.

0:16:49 - (Micah): All right, then go to ClickUp and go, well, where's Client X's folder? Okay, cool. Where's their backlog? And then create the task like it's not that hard, but that takes time and it's distraction. And when those warnings and errors come in, you want to act quickly. You don't want to, oh, our PMs were in a meeting and so we didn't even create the task for four hours. But an agent is just sitting there and it takes three to 10 seconds because we can say to the agent, watch this email, read the email, find the right client in ClickUp, find their backlog, check to make sure there's no other tasks that relate to this and if there's not, make a new one. Otherwise add a comment to the task that already exists.

0:17:44 - (Micah): I don't know how you would ever program that.

0:17:48 - (Alane): It would be so complicated. There's too many different scenarios that it could fall in for programming. But you know, if you take this idea, Micah, that you just said, we're using a very specific type of email coming in where we're monitoring errors and automation and looking at those. But it could be anything at your company where you're getting client requests, you're getting client questions, you're needing to look things up that a client sent over, you know, whatever that might be. This same scenario can be ran through where it's basically triaging the emails, reading it, creating a task, it could even draft responses for you. And again, we still need the human element.

0:18:30 - (Alane): You still want to read it before it sends and, but the time that that saves of going back and forth, getting things ready for a team member to review, because as an account manager, you may not know all the answers to everything. You might need an internal person that goes in and does that. And now that part can get to the team faster.

0:18:50 - (Micah): Yeah, yeah. And it's great because we, you know, with our team, our project managers are going, oh, this is awesome. I love this so much. Because now they have daily alerts about the clients, they have all the error handling emails taken care of for them. The tasks are automatically being generated. We're looking at this going. We're getting automatic reports that we don't have to do from a monthly and quarterly level.

0:19:19 - (Micah): Like when you start thinking about all the different areas of a business to keep it running efficiently and to supply your team with the information that they need to do what they need to do. That's a no brainer.

0:19:33 - (Alane): Yeah. So let's talk about a couple of, I don't want to say cons, I'M going to say a couple of opportunities that companies, that, companies that want to use AI agents in, in all of this. So the episode before this is about standardization so that you can scale and this is still a core piece, the standardization piece. Because if you have chaos, the AI agents is, they're going to be digging through chaos and will not be successful.

0:20:03 - (Alane): So having that piece of standardization so that you can use AI AI agents to automate is still a top priority.

0:20:13 - (Micah): Yes, that's a great point.

0:20:15 - (Alane): So if you're wanting to do this and you're looking at, well, you know what standardization do I need, even naming conventions for files, having a clear structure for that, even that is a basic need for something. Because if you're saying, hey, I need you to look up this, this and this, it needs to know what client and what it is that it's looking up. And that needs to be in the file name.

0:20:38 - (Micah): Yeah, I think what you're, what you're bringing up, Alane, is, you know, there's an old saying in, in it and computers is garbage and garbage out. And it basically means like if you're going to feed it garbage, you're going to get garbage as an output too. And it's still to, to, to a degree or for the most part, the same thing with AI agents. The good part is we don't have to have like we did even two or three years ago, like we had to have form fields with specific drop down options so that we could write logic against that.

0:21:15 - (Micah): We don't have to have that anymore. So that's always a nice, a nice benefit. But yeah, organization is key. I think kind of a rule of thumb is for this, if you were to hire somebody brand new and you were to ask them to do something, if they couldn't figure out what the answer is, chances are an AI agent isn't going to magically be able to read your mind and know what it should pick or, or should do or how it should, you know, respond to something. It has to be structured enough like you're saying, so that it can put the pieces together and go, yeah, okay, yeah, I think this is what the user's asking for.

0:21:57 - (Alane): Yeah. And you know, to clarify, Micah, what you just said is if you just hired a new person and told them to go do this, they can't ask you questions because you go, well, you know, like, right, I could do this, this and this. Like, no, then you just trained somebody and so they could ask you questions all day when they can't find something. So you need to think about it that you just hired somebody and they cannot ask you anything. You assign them something and they need to know that they can go do it without any clarification. That's your, the goal that you're trying to achieve.

0:22:29 - (Micah): Yes.

0:22:29 - (Alane): The other thing that I think is, is an opportunity is that you think about these things as still an automated workflow flow. It it's like, you know, you incorporate an AI agent into an automation. It isn't just the AI standing AI agent standing alone. It is still an automated workflow. So it's piecing together things. It could be between Gmail, AI Agent and ClickUp or something like that where it's still, you're piecing together multiple software platforms. So when you're thinking about hey I really want to be using AI agents, even in our toggle scenario, we're using toggle AI agents and Slack like we've got an automated workflow that it is, we are telling it and part of that automation. So it's not just you have this AI agent, this one little thing and it goes and does a bunch of things. When you look at it, it actually looks a lot like you would in make where you've, you're telling it a, a number of steps to do.

0:23:31 - (Micah): Yep. Yeah, excellent, excellent point. So here's my prediction on AI agents.

0:23:37 - (Alane): Let me hear.

0:23:37 - (Micah): Regardless of what it ends up being called, right? Who knows if agentic workflows are gonna stick or AI agents or whatever it ends up being called. I don't think in five years or less there's going to be a business on this planet that can compete without having some form of AI agent involved in its workflows. Everybody else that's using them is going to be faster, smoother, quicker and be editing, editing, editing instead of creating from scratch, creating from scratch, creating from scratch.

0:24:15 - (Micah): It's a big difference and it's a game changer. That's my prediction.

0:24:20 - (Alane): I think, I think that's a good spot on prediction Micah and you even being a customer of other companies, I was just telling you last week how frustrated I was. I could tell it was a small mom and pop organization that did not value software because the process was slow, time consuming and a burden on me because it was taking my time and do I want to do business with a company like that, I will in that scenario. I already started it. But return business doesn't happen that way.

0:24:54 - (Micah): Because I mean it was a donut shop, Alane.

0:24:57 - (Alane): So it was not, it was not a donut shop. But you think about the cool things that make life easier right now. And it is the companies that are doing this because for their customers, it makes it easier to work with them. So be thinking about that, listeners, as you evolve your processes and you're ready for AI agents and using software, is that you are making it easier for customers to work with you.

0:25:23 - (Alane): There's one more thing that I want to say to close this conversation out, Micah. And because I get this every single time I bring up that we do AI and automation and then the person says, oh, are you doing anything with AI agents? And I say, yes, we are. We're doing all this cool stuff. And it always happens that they say, oh, you know what? I've been seeing that some of the AI agent companies offer free AI agent setups.

0:25:49 - (Micah): Yeah, you're talking about the platforms.

0:25:51 - (Alane): Yes, the platforms themselves. And so there's nothing wrong with that. But I think the misconception that people are having is, oh, I'm gonna sign up for this account, they're gonna help me build up these AI agents and my business is going to run so smoothly without people. And what the, what they are setting up for you are basic AI agent workflows. The very templated, really just setting up.

0:26:19 - (Micah): The, just the framework of the agent.

0:26:23 - (Alane): Right.

0:26:25 - (Micah): It's not the details that makes them.

0:26:29 - (Alane): Actually flow and it's the details that make these powerful for your company. Without it, it's like, yeah, that's kind of cool. So, you know, definitely, you're nothing to do with.

0:26:41 - (Micah): I know this was the last thing you wanted to say, Alane, but I do want to piggyback on that for just a second because I think it's really important. For example, when it comes to AI automation, when it comes to AI agents, there is a huge amount of nuance for optimization. So for example, our first version of the time tracking agent that we built. Right, right. We gave everything to the agent. We said, do all this, process this, do all this.

0:27:11 - (Micah): Our expense for having an AI do that was huge because it was such a large amount of data and then it was not quite predictable enough because we're asking a language model to do some math. So the workflow around this or the work, the, the solution, the optimization around this is to say, well, great, let's pull all the data first. Let's then write some code to process the data in a way that we know will always be predictable. It will always output the same way.

0:27:46 - (Micah): That's the power of, of programming and code and pulling from APIs. Like we always know it's going to work when the APIs are up, but when we run it to the programming language, it's always going to produce the same output. IT LLMs and AI agents are not that way. So what we did was we, we pulled all the data out, we ran it through script and then we give the, the condensed version, only the metrics that the agent needs to work with to the agent and then asked it to process that.

0:28:24 - (Micah): So our token count was like 1% what it would have been if we just said, oh, let's set this up and like here's, here's the API access agent and here's all these instructions and go do this just didn't work very well. But when we optimize it now we're, now we're in the right area.

0:28:46 - (Alane): Yeah, I mean Micah, you, you really hit on a lot of things. And, and I was thinking through too, there is a technical aspect to this that, that you need to, to be, to be a strong AI agent workflow builder. And there is another side to this. When people are trying too hard to custom write this stuff to where it takes so long to build an AI agent. So you want to have the balance of utilizing technology to make this a quicker build and you need to have some technical acumen as well, which you and your, you and our team do.

0:29:21 - (Alane): Where. That's why I'm the strategic idea creator on these.

0:29:25 - (Micah): I mean this agent was your idea, the one that we're talking about.

0:29:30 - (Alane): So I love great ideas. Just I need some people to help me execute on them. And that's, you know, what we do for our clients too is they've got ideas of things that they want to do and we're here to help on building those AI agents the best that they can be and do. All the cool things that Micah just explained blamed on a optimized AI agent workflow. So if you want to talk AI AI agents more, if you want more information, reach out to us. We obviously are very passionate about this piece of our company and this growing opportunity that we're seeing.

0:30:04 - (Alane): 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. 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 automating your business. Join us next week as we dive into more ways to automate and scale your business.

0:30:30 - (Alane): Bye for now.

People on this episode

Head Shot

Alane Boyd

Co-host
Head Shot

Micah Johnson

Co-host