S1. Ep7 - AI Chatbots

Katie: Welcome to the AI Automations for Business Podcasts. We're your hosts, Katie and Noel, 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 and I hope you enjoy this episode. Hello, welcome back to another episode with us. Hello, I'm Katie.

Noel: Hello, and I'm Noel.

Katie: This week, we are talking all about chatbots. So for anyone listening and you're thinking, I like the sound of this, but actually, what is it? Noel, can you just give us a really brief overview of what chatbots are?

Noel: So a chatbot is something that you can host either on your website or within a particular application that you use data day to day. And what you can do is then link that into different tools and AI and then basically talk with your data. that's kind of the up shop and also do different things yeah

Katie: and most of us have probably experienced chatbots when we've been on someone's website like for customer service you know you you see it pop up don't you usually in the right hand corner like how can I help that is a chat

Noel: box absolutely yeah and the thing with those sort of chatbox is like most people might look at those and think well that's either going to be really complicated for me to incorporate into my business or really expensive And, well, I've had a quick test around last week, and it's not either of those. It's cheap and easy.

Katie: Okay. Let's get into that a bit more. But first of all, it's like, so we kind of now understand like what they are, but how can they really help our businesses? And I guess, like, first of all, it would be really great to understand what the different type of functions you can do with these. chatbot's first.

Noel: So I think the big plus is if you had a chatbot on your website, you could always be there. So they're not filling out a contact form and they're maybe getting a response in eight hours.

Katie: Yeah.

Noel: They're getting information live, especially like support and things like they're getting help they need when they need it. They're not having to sit around and wait for someone to respond. So that's kind of the real big plus. And yeah, there's all kinds of different things and functions that they can do, and that kind of depends on the aim of the chatbot that you would like to create. So yeah, there's all sorts. I mean, I guess the common one would be to have all of your data within the chatbots. Anyone asks a question, you know, so let's say for Business Wonderland, you know, for you, Katie, you know, you could have all of your content in there, and then people could ask questions and then get responses back as if it was you.

Katie: So this would be really great as well if someone's perhaps running a community. or a membership, you know, someone can't find something or they want help with a certain subject or topic and they can ask the chatbot and rather than you having a customer service person on your team, answer it, the chatbot would be able to answer it within seconds.

Noel: Yes, so I built a chatbot for the AI community on the school platform and basically they had like thousands of posts, loads of YouTube videos. So there was lots of really good information. But what new users were finding was it was just like so overwhelming. They couldn't find the answer to the things they wanted. So, you know, we created a chatbot within the DMs that people could message. And then it had access to this massive knowledge base. And the AI could just go through it and go, well, you know, they're asking about this particular tool. Here's the answer. And here's the video that you need to go and watch that's within the community. and it will go out, you know, pinpoint people in where to go.

Katie: Yeah, so it wasn't just coming up with one particular thing. It was saying, you know, check this out or check this video out. And you also might like this video as well, for example.

Noel: Yes. Yeah, it was giving a, we kind of restricted it to like three responses or three things it would kick out. Just so.

Katie: So it wasn't more overwhelmed.

Noel: Yeah, exactly. Because if you just asked about May.com, there's like 70 videos just on that one. So that's not particularly helpful.

Katie: No, not helpful at all. Yeah, and we want to reduce that overwhelm. And I think that's the thing. If you're running a community or a membership, people do get overwhelmed very quickly with all the information or even knowing where to start. So something like this when, you know, you've got people, you know, a community from all over the world and perhaps, you know, your community manager or customer a service team, you know, only work particular, you know, set of hours. You haven't got a global team where someone's working all the time. The chatbot would be a great alternative.

Noel: Yeah. We also saw like a really big downturn in people asking questions as well. So they were going off and using that chatbot to, you know, and getting the answers they needed. And even if they ended up creating a post asking a question, like me and the other admins would then go off and talk to this chatbot and then say hey what you know can you answer this question for us and it would be like oh yeah you need to look in this video this video right you know and then then answer the answer the posts so even we're using it internally

Katie: so if someone was asking a question in the community rather than asking the chatbot you could then copy their question into the chat bot and then get get the response and then paste it back into the community.

Noel: Yeah. I know there's some community members listen to this podcast. So yeah, I do apologise if I've answered the question. But yeah, I've been using the chat bot to for all the answers.

Katie: And would you say they're like 100% accurate? Because obviously, you don't want to be telling your clients or your community members wrong information. So that would be my concern.

Noel: Yeah. And that's what. kind of the worry that we had when we first started it. It went through a hell of lot of testing just to make sure. The thing, like we were saying, you can connect multiple or, you know, or one tool to a chatbot. So for this particular chatbot, we had what we call a vector store. And that's kind of like a fancy database. And basically, it allows you to input all the data and it kind of puts in like a 4D grid. It's really, yeah, you kind of need some sort of sign degree to really understand it. But what it does is it pairs like words together within the database. So every time like in a transcript someone says make.com, you know, it would then link all of the other video transcripts that said make.com to that, you know, word or phrase. So then when you ask a question about it, it will query the database like you would do like a Google search. And yeah, then it comes back and it says, well, look, all of these had make.com within it. And, you know, this one had the best answer overall because it's gone off and it's read all of the data for each of those inputs. It's quite complicated, I must say, but it's really good to how it works and it gives you really accurate responses. Okay.

Katie: That's all we needed to know. Accurate responses. So we said like people with communities or memberships, they could use it. I know you briefly said, like, you know, you could put it on my website. So for those who who don't know me, I'm a business and marketing coach. So I offer obviously coaching and consulting and I've got courses. I've got my own podcast as well with a blog. So you're saying that I would be able to use it. But what other business do you think would benefit from a chatbot?

Noel: So anyone that's like selling products that you always have like FAQs or, you know, or need to provide support for, that that's really handy because you can store all of that information in the vector database. And then if anyone ever asked a question about a particular product, the AI can go off and then go, well, tell me about the product. And it will come back and then give you the answers that you're looking for. And also go through that. Yeah.

Katie: Yeah. So for example, if you're running a, let's just say a skincare product. So you've got a line of different skin care products. And that's, They're all four different skin types, so one for dry, one for sensitive, one for oily, one for ageing, you know. So people might not necessarily know which skin product is going to be the best for them. So then you could have, you know, a chat bot to answer questions like that, like which skincare product is the right one for me.

Noel: Yes, yeah, because it would pick up on words like oily skin and moisturiser and it would use those within like a keyword search. and then it would then, let's say that then brings back five products. The AI would then read all of the information about those five products and then have another look at the question that came in and then pair you with whichever product would work best. It could be all five. It could just be one or none at all. But at least it would come back and then say, well, look, you know, we've looked at this. This might be an option, you know. And then you could then even import like URLs into the database. So, you know, when the response comes back through, there's a link that you can click on and it will take it to the product page.

Katie: Or, you know, if the chatbot doesn't understand it, just go from what you've described. We think it's best that you talk with one of our customer service representatives to ensure that, you know, we make sure you get the right product sort of thing. Yes.

Noel: Amazing.

Katie: Love that.

Noel: Yeah. Yeah, it's really, really easy to do as all.

Katie: Can every beauty website have that on, please? Because that would save me so much time.

Noel: Yeah. Yeah. I think what? Yeah, I think like most companies do, don't they? But I don't think they go kind of to that level.

Katie: No, they don't. No,

Noel: they don't. Sometimes you use a chatbot and you think, well, what's the point of even doing this?

Katie: I know some chatbots are so terrible. So I was using one the other day and like I put in the most simple basic command and it couldn't like answer me in. So it had to connect with, like, actually had to connect me with, like, a member of their customer service team. It's so annoying. I was like, yeah, from obviously the work that we do in AI automations, I was just like, you should be able to have done that with your chat. It's so annoying.

Noel: Yeah. But the way that they've probably built it, it's probably with, like, really dated tools that we would see now. So what we would see is dated now. But yeah, I think now it's like really easy to. inject AI into these sort of chatbots to give you that level of intelligence, you know, to answer those sort of questions and have access to millions of data points for you. Because you, I mean, Amazon could do it, but that'd be a gigantic database.

Katie: Oh my gosh, can you imagine? But the thing is, though, with Amazon, like their search function is actually really decent.

Noel: Yes.

Katie: So if someone did want to build a chatbot, how, like, how are we going about doing it as in, is it going to be costly, you know, how much time is it going to take?

Noel: So the one I was looking at last week, so there are, I must say, there are several tools out there. There's things like Bot Press as an example, which is a paid service. But the one I was using was my old friend, N-A-N, and what that allows you to do is you can create an account online and then use their cloud service, and that's $20 a month. And then you can, you know, make some workflows within their platform. And what they have is like a really handy chat input function. Now, I just thought within ANAN, when you connect that in, you can just like, you get like a little chat window and you can start talking to the AI. But I thought that was it. I didn't think it could go any further. And then something random popped up on YouTube and went, oh, well, no, no, no. You can put that on your website. I was like, you can't. No way. And it was really simple. And it was like, yeah, you just, you know, you just. toggle switch and then there was a particular help page within the docks and it was like yeah this is the structure of the code that you then copy paste into your website and that's it it's done it's all connected but i connected to it within like 10 minutes so yeah it was really straightforward to do but the next level to that is like you're you're then paying $20 a month yeah but what you could do with n-at-n is host that locally so you could create like a virtual server and install NAN locally on there. So like a virtual server is like four, four, five dollars a month. They're not really a great deal of money. So, you know, you could then host it on there and then use, you know, NAN to your heart's content. There's no usage limit like there is with the cloud version. You know, so if you've got a chat bot that has 25,000 people used it a day, you know, a thousand operations or, you know, things that you, I think they call them operations in NAN, But you're going to blitz through that within the first 10 minutes. So, you know, hosting it locally is like really, one really cheap. And it just opens up and you can, you know, you can use it as much as you like or as little as you like.

Katie: Okay, so budget friendly.

Noel: Very.

Katie: And then what about time to set it up? Because obviously you've set up a couple of these now. And I know one in particular did, you know, it was quite time consuming, wasn't it?

Noel: Yes. Yes. So it depends on like the platform that you want to use. So the one I did for school took me a long time. But then the NAN that you can put on your website, not very long at all. But what you need to do is, I would say, is to figure out what the key function is and then keep to that key function of the chatbot. But don't try and create a chatbot that does absolutely everything. You know, just try and keep it. You know, so if it's a customer support one, you know, that only pops up on your customer support page. And it's dedicated to that particular function.

Katie: Yeah.

Noel: You could have one on like a products page and that only does products.

Katie: Okay, so you can have several chatbots on the same website.

Noel: Yes.

Katie: My gosh. Okay.

Noel: Yeah.

Katie: I did not know. Yeah.

Noel: I didn't realize I could do it until I give it a go. I thought, I wonder.

Katie: Yeah.

Noel: Yeah. You could have all sorts. Okay.

Katie: So yeah. So one chatbot for one main function.

Noel: Yes. Yeah. It makes life so much easier. So you could use like an AI agent to handle multiple tools. But in our previous episode, we went through maybe that's probably not such a great idea right now because it's not.

Katie: Just because of the accuracy at the moment.

Noel: Yes. Yes. Yeah, as of right now. But yeah, it's getting, it will get better over time and it will be a lot more accurate. But yeah, for now it's kind of, you know, one function at a time. And then, yeah, that's, makes life so much easier. But if you're doing things like a product database and things that and then having people having access to ask questions about beauty products, like we're saying, you know, that you need to really think about the structure and what you're inputting into that vector store. And then, yeah, then that allows the AI just to, you know, they then understands the structure because it's the same every time.

Katie: Yeah.

Noel: And, yeah, it just makes life so much easier. Yeah. Okay. Amazing.

Katie: Well, thank you so much, Noel, for talking us through chatbots. If anyone has any questions about chatbots or, you know, you want to set up your own, but you're not sure how then you can find us in our free LinkedIn group, which is AI Automations for Business. I'll pop the link below for everyone. But yes, feel free to come and ask any questions in there. Thank you so much, Noel, for All of your amazing advice, as always.

Noel: Not a problem. Glad to share it.

Katie: If you've liked this episode, then we would absolutely love for you to subscribe. And if you feel cool to leave us a review. And we will catch you next week for another episode.