S1. Ep14 - AI Agents vs Automations: What Does Your Business Really Need?

Katie: Welcome to the AI Automations for Business Podcasts. We're your hosts, Katie and Null, and we'll be discussing how you can use AI for your business along with the latest news, updates and automations to help you stay ahead of the curve, allowing you to grow and scale your business more efficiently. Please be sure to subscribe

Noel: and I hope you enjoy this episode.

Katie: Hello, welcome back to another episode. Hi, I'm Katie. And I've got Noel with me as well. Hi, Noel.

Noel: Hello.

Katie: We are talking about AI automations and AI agents this week. So we've seen online that there are a lot of people getting confused with what an AI agent does and what AI automations do and people thinking that you just need an AI agent. But actually sometimes you just need an automation, right Noel? Absolutely,

Noel: yeah. There's a lot of people also selling agents, but actually it's just an automation. Okay.

Katie: So we're just going to go through what the difference is and then why you might want to use an AI agent versus using an automation. How's that sound?

Noel: Sounds good to me.

Katie: Okay. Where do you want to start?

Noel: I guess we need to start with defining what an automation is. and what an agent is, that's always a good place.

Katie: Okay, so let's start with what, let's start with what an AI agent is.

Noel: So an AI agent is something that you can talk with. So you can have a conversation with it and say, well, look, I would like you to do this complicated task and you can ask it all in plain language. But within that plain language, you might have asked for three different things. And then the agent, what that does is it looks at all the tools it has available. and then it goes, right, okay, well, how am I going to do this task? What do I need to do first and how am I going to execute it to give me the user what I need? So that's kind of an agent, it makes things up. It's got its own brain and it will figure things out for you. Whether or not it does it in the right order, who knows? What comes down to prompt engineering?

Katie: Okay, that's good to know.

Noel: Yeah.

Katie: Okay, and then, so what is an AI automation?

Noel: So an automation is something that has a defined route. So as an example, if you have like a social media planner, you might say in one column, you might say, I want a post that is about AI agents. And then you might then select a column that says, I would want that on Instagram. And then because you've created that selection, within the automation, it has like a defined route. So it would say, well, you want a post and it's Instagram. So it goes down that defined route. and you get exactly back what you ask for. But in an automation, there's no conversation, so you can't have a chat with it. And say, well, actually, what you've done isn't quite what I wanted. You've kind of got to roll the dice again and hope you get what you need. Yeah.

Katie: So with an automation, it's kind of like you know where you want to go. You know the output that you desire, where it is an AI agent. It's more like a search function. Yes.

Noel: Yes, kind of, yeah. Yeah.

Katie: I know it can do so much more, but...

Noel: I love a lot more.

Katie: Yeah.

Noel: But like, so the minute I have a course on Udeme that does like a social media automation, so you define those paths in Airtable, that it creates the post. And I'm now converting that into an agent. So like with an agent, I've got like said like, there's the search function so I can go off and say, well, could you research this topic? And it will come back and go, well, here's all the trending things about this topic and I'll go, right, create a post and an image and then it then looks, it understands the conversation I'm having with it and then goes, right, okay, well, we need to create a LinkedIn post, so I'm going to use the LinkedIn tool, and then it needs an image, so I'm going to use the image tool, and then combines all that together and gives it back. But if I don't like it, I can always go back to me and say, well, actually, this isn't quite the angle I was looking for. Could we change it and then it will go back and then reuse those tools to get the answer that you need? Yeah. Really quite clever how it works. Yeah.

Katie: So where would you use an AI agent and where would you use an AI automation just for, like, an example so people can kind of like get a good understanding?

Noel: So with an agent, I would say you could, you would use that at any point where you either need to have a conversation with the AI about whatever tasks you need to do.

Katie: So just to pause there.

Noel: Yeah. Why?

Katie: So people might be thinking, well, I can do that with something like Deepseek or ChatGPT.

Noel: Mm-hmm. That's true. Yes.

Katie: So what is the difference then from using something like Claude, ChatGPT, Deepseek to using an AI agent?

Noel: So what you can do is you can give it access to all of the tools that you currently use. So my old friends, make.com, they've got their own agent. So we can create tools using all of their like 2,000 apps that we have access to. So you could give it access to your CRM and things out and say, well, look, if these are all my clients, could you find one that I should probably follow up with, you know, and then create an email. And that could be two different tools. Whereas in ChatGBT, BT, you could do that, but you would have to go off in your CRM, have a look, find them, and then provide, copy, paste all the context, a big prompt, and then go, right, create me an email, and then you get it back, you're like, well, that doesn't really work for me. Whereas all of this altogether, you know, it would look at your CRM, create emails, and then also put them in your draft folder, ready to send kind of thing, so you don't really have to

Katie: do anything.

Noel: It's more like an assistant,

Katie: isn't it?

Noel: Very much so, yes. Yes, a very powerful one.

Katie: Very powerful one. No, ha, ha, ha.

Noel: Yeah, yeah, got to be careful how you use it. Okay,

Katie: so let's have a look at then some pros and cons. of using AI agents and AI automations. So what one do you want to start with this?

Noel: So with an automation, and we'll start with those ones, is if you've got like a predefined process, so you're creating the same type of document or you're creating the same sort of content each time and you know you have predefined routes that you want to use, then an automation is the only way to go on that one. So the big pro is it will always go down the routes and use the tools and applications that you've connected in the automation. It won't go off and make things up. So it's very consistent. And for businesses, that is ideal. A lot of people kind of want the same sort of stuff every single time.

Katie: Yeah.

Noel: So those

Katie: repetitive tasks that most of us have within our business.

Noel: Yes. And the con for an agent on those sort of processes, you could give access to all of the same tools, would in an automation. But every now and then, that agent might decide, you know what, I'm not going to bother using that tool. I'm going to do something different and then, you know, and then use something else to say it's a similar output. So it's not like a pre, it's not like hard coded in. So it's very flexible, which isn't ideal for some people. Yeah.

Katie: When you know exactly what you want it to create week in, week out, for example, you want it to have the same result. whereas an AI agent because it's almost got like a mind of its own.

Noel: It does.

Katie: Mike, do you know what, I'm not feeling that this week. I'm going to tweak it or I'm going to skip a step or, you know, I'm going to give you a completely different answer.

Noel: Yeah. So what I was finding with one client, so I had a call with them and she said, oh, let's book in another call. And she said, oh, but I've just built this agent that will do all this for me. and she sent a voice note to the agent and said, you know, booking this call with Noel, this is the time, this is his email. And it came back and said, yeah, all done. It's all booked. And we were both looking at our phones and went, I haven't had anything of you. And she was like, no. And then we looked at the automation. And basically the agent had just gone, I can see I've got access to Gmail and Google Calendar, but let's just say it's done. So that was one particular tool. So make.com's agent is a lot better. It will actually use the tools. that particular one not so great easy to build and that was with n8n so that was easy to build but yeah just kind of ignored what tools you give access to which isn't ideal because obviously if you're wanting to send that to a client you know and say well you know let's book in a call and you're like yeah great get the agent to do it and then it comes back says yeah I've done it you know okay but did it

Katie: so for that case would actually an AI automation have been a better use rather than the agent Yes.

Noel: In a way, yes. Yeah, yeah, you could have, there would have been ways to do that.

Katie: So I'll let you keep going with your pros and cons.

Noel: Pros and cons. So I guess a real pro when it comes to agents is like the ability to talk to it and then amend and things like that. So that could be really handy. So I use Telegram to chat with agents, but you could put that into any chat thing that you currently use. It could be Slack. It could be Microsoft Teams. You know, we can connect all of that information in there. and then just have a plain conversation. Because what I find with some of the more enterprise apps and agents is that you've kind of got to put in like keywords. And if it doesn't understand the keywords, you don't get the response where there's the ones we can build within like make.com and things that it's a lot easier just to have a plain conversation. Yeah.

Katie: That makes sense. Yeah.

Noel: So I did show it off to a recruitment agent. And they were like, well, we've got an agent and we've got an agent and we've give it all these key words, but if we haven't got it right, you know, it doesn't do this. And I said, well, let me get my telegram bot out with, and I got all these CVs and stuff ready. And I said, hey, could you find me someone in Bristol that's a Python developer with like 10 years experience? And he went, yeah, sure, here's all the CVs for those people. And I was like, right, brilliant. Send me, send an email out to Jeff.

Katie: And then off it went.

Noel: And it created an email and put it in my inbox, and my draft box. And they were just like, blown away. They're like, I can't believe he just did that with, there was no keywords, you know, you just asked like a question like you would do it to anyone normal employee.

Katie: Yeah, which is really cool.

Noel: Yeah, it is. Yeah. Yeah, it's amazing. Yeah.

Katie: But it's just about making sure that you are using it in the right capacity, isn't it?

Noel: Yes, yeah. If you're happy for it to go off and do its own thing, then yeah, agents are a good way of doing it. Yeah.

Katie: Because I feel as well, like there's such a buzz around AI agents that everyone thinks that they should be using them. Would you agree?

Noel: 100% yeah. So I had a chat with a potential customer yesterday and they were like, I want an agent. And I was like, okay, what do you want an agent to do? And they were like, are always going to do like these managed whatever processes? You don't really need, you need an automation, not an agent. I was like, you know, we can figure that out. And they didn't, I don't think they really understood the difference between the two. They just knew that agent was, the latest buzzword and it was like, I want that. And it's like, well, yeah, you can still have AI in that process, but maybe not an agent because it's just going to, it might not do what you want it to do.

Katie: Yeah. So yeah.

Noel: But I also find it on platforms like Udeme. So I was doing some research before doing my course. And then I was, I came across a course that was almost identical to what I was doing. I was like, oh no, I've got direct competition. I looked at it. And I thought, oh, I'll just build an automation. I think he was calling like the AI modules within his workflow agent employees. And I'm like, well, that's not true. They're not agents. That's just an AI that's there to answer and give a specific response to a specific part. That's an automation. Yeah.

Katie: And as always, like we always say come and join us in our free LinkedIn group, which is AI Automations for Business. And if, you know, these are questions that you can ask us, you know, you know, I want to build this and you commit, you know, you can ask us, you know, would an AI agent be better or would I need an AI automation?

Noel: And, you know,

Katie: we can come and help you with that. We will leave the link in the show notes. But if you just search on LinkedIn, AI Automations for Business Group, we should pop up. It's completely free to join. Everyone's welcome to join. You know, and you can always come and show off what you're building. with AI automations or, you know, if you've built an agent, come and tell us because we absolutely love to hear what you guys are up to. I know one of our members, she was telling us that her daughter runs an Etsy store and she was using AI automation for that. And I just thought that was so, so cool. Like, I just love, you know, learning what you guys are using AI automations for. But yeah, do come in and use that group. It's there. It's a lovely community. It's very supportive. And yeah, we put a lot of updates in there as well, like when new things get released and things like that.

Noel: Noel, anything else to

Katie: add on AI automations and AI agents.

Noel: I think we've covered all the basics just to arm you with that knowledge so you know the difference between the two. Yeah.

Katie: Knowledge is power, right?

Noel: Absolutely, yeah. Yeah. Don't buying an agent when it's an automation.

Katie: Yeah. Don't get caught up in the hype.

Noel: No, definitely not, no. There's a lot of it.

Katie: Yeah.

Noel: This isn't it.

Katie: Amazing. Well, thank you so, so much for listening to this episode. If you like it, please come and leave a review for us. It really helps us so it spreads the word. So more people can learn more about AI Automation's business. and we will catch you on another episode very soon.