Episode 18 From L-Plates to AI Boss

Untangling Tourism Tech Episode 18, with Fabienne Wintle and Liz Ward. Each line shows who is speaking and when. Times match the video. We removed filler words and fixed mis-heard names.

[00:00] Liz Ward: Hey Fab.

[00:01] Fabienne Wintle: Hello, Liz. Wow, you’re looking great today.

[00:04] Liz Ward: Would that be because I played with my appearance in Google Meet just to do a little makeover?

[00:11] Fabienne Wintle: Isn’t that so good? We’re just too busy to actually get that makeover done, so we get AI to do it for us. But hey, if you want to jump in, in the client’s meeting, you’re going to immediately look a million dollars.

[00:24] Liz Ward: But this is definitely us and not our avatars. And I would very much like to welcome all of our listeners and particularly our Tourism Tribe members along to this podcast of… this episode of Untangling Tourism Tech. And this is Episode 18.

[00:44] So today, Fab, we’re going to take a bit of a journey through what it looks like to progress and build your confidence using AI tools. We’re going to have a chat about where the industry is at presently, what that journey, that road map can look like in developing those skills and confidence. And then we’ve got a really great conversation to have about a real-life example of a tourism operator trying to untangle a problem and how AI could be used to do that to build their efficiency and get rid of some workload that they recognise is really a pain in the neck for them. So stay with us to get through to the live examples.

[01:29] And let’s have a chat first of all about where the industry is at. So Fab, when you’re coaching either destination marketing organisation staff, council staff, operators, what are you hearing at the grassroots in terms of where they’re at with using AI in their organisations and their businesses?

[01:49] Fabienne Wintle: I feel like they’ve plateaued with using tools. There’s a lot of mention of ChatGPT, quite a few mentioning Gemini as well, but using the same prompts, the same systems, and knowing there’s way more that can be done, but not knowing how to go about it. It might be the AI, the tech block, you know, like how do I break this down in steps? But a lot of people in organisations, whether they’re councils, tourism organisations, are using ChatGPT or whatever tool is allowed in their organisation. And there’s a lot of tools that are blocked, and we hear a lot of people having to use their own account or work from home. A bit like when, you know, organisations were not letting their staff enable Facebook pages. I remember all those years ago. So I feel that’s where we’re at. So the system needs to catch up so they can actually start being really efficient and productive.

[02:51] Liz Ward: Yeah, I think there’s a real human factor here too. When I’ve been out on the road recently doing face-to-face workshops, or when we’re doing online workshops, and we ask the question about where are you with your usage of AI yourself personally, what I’m finding in the regional tourism area is we see 80% of hands go up that they are experiencing AI, and it is for… they’re kind of on that L-plate. They are using it for creating content, maybe a bit of research, and they feel that that is fantastic. They’re like, “This is amazing. I can get this done so quickly,” which is great, but they don’t know what is possible beyond that.

[03:40] And when you ask them what tool are they using, I would say 80% of those people are using ChatGPT, and then you’ve got some outliers who are loving Claude or Perplexity or another LLM chat tool to get those sorts of jobs done.

[03:59] So it’s great that the industry has gotten there. So we’ve got 80% who’ve had a play or are using something like ChatGPT, but then you’ve got 800 million plus, you know, queries a week for using something like ChatGPT by the consumer to be able to do travel planning. So, as you say, the industry is trying to catch up, and I think where the difficulties come is having the time to get in and play and experiment. And that’s certainly been, I think, our experience, hasn’t it? That you’ve got to get in and use the tools and then you build your confidence and you can work out which ones you’re going to pay for and keep and how you really apply them in the business. And you’ve generally got that issue with small businesses who are time poor and don’t have the luxury of that.

[04:52] So where we’ve tried to help people over the last few months is to think of it as taking the analogy of getting a full driver’s licence. So I would say I was on my L-plates last year, where I was loving ChatGPT, doing lots with it, like really pushing it and getting the most out of it, which was fantastic. And I started to progress to my P-plates, and I’d say you’re definitely on your open licence, Fabi, and we’ll talk about that in a sec. But for me, that journey was about moving off the L-plates, doing things like developing a custom GPT. And a lot of tourism people in business don’t even know what a custom GPT, or what the equivalent is in Gemini, a Gem.

[05:40] So, a preloaded set of instructions that allow you to do repeat tasks, whether it is creating social media posts or writing really well crafted and tailored review responses, developing a marketing plan like our Tourism Marketing Mate custom GPT you can find. So that’s an example of being able to grow those skills and move off your L-plates and realise that there’s more that you can do. I’d say moving into your P-plates, it’s also using tools like Gemini, their advancements, Gemini’s advancement and generally Google’s tools advancements in being able to produce things like… produce amazing re-versions of your existing images. So, doing editing of your existing images, and I know you’ve got a great example of using Gemini to do that for a client.

[06:42] Fabienne Wintle: Well, so many clients, you know, it’s so good. You can literally dump an image. I was working with a restaurant and they had pictures of the tables outside with one person at it. And we all know in tourism, like, you actually want to… displaying a bed with no one on it is no good. You want to get people experiencing your product. So ages ago, I just tried it and I said, “Can you just put some people in there?” That was literally my prompt. And it just set people at the table, and they look like real people. No AI images, you know, like you might still see some AI images on Facebook people are doing, but we’re not there anymore. People look real. They’ve got, like, you know, a 45-year-old face. They’re not all polished. You feel that they are really real people, you know, there’s a dog bowl there, everything’s good.

[07:31] Even if you have your pictures, like another client of ours, you know, they sell merch and they would put their hats, pictures of their hats, on their website. I’m like, “Actually, let’s just put a person wearing the hat.” And bam, immediately, with the name of the client, Gemini produced that photo. So there’s no excuse not to learn how to do that. You don’t need to be a tech person. You just need to think, “Oh, well, let’s try with an image instead of text.”

[07:58] Liz Ward: And I think the other one that’s really applicable is AI in video as well. There are so many tools that allow you to quickly now create good video, particularly for social media purposes, where the appetite by users of TikTok and Instagram and even Facebook for short videos has just grown enormously, and we operators and destination marketing organisations have to be creating that really authentic video content. So the tools available to create videos are just… they’ve moved so quickly. And I know, Fab, you’ve got a couple of favourites. Do you want to just talk about those, so our audience knows where you’re playing at the moment?

[08:47] Fabienne Wintle: Well, I got really excited last weekend because the way I learn right now, it’s quite interesting. People think I read books and things. No, I actually watch TikToks. I watch short TikToks about AI and what I should be doing, and that’s when my brain is like fried at night. But I’m like, “I’m just going to see what’s been happening in development.” And I’m like, “Ooh, Remotion, what’s this?” And it’s actually a tool that allows you to create video using code. So that’s really exciting, because by definition, video, you need a user interface to move things. We all know, like CapCut, etc. I’m like, “Ooh, can I throw the blog post in my automation tool that I’ve created, and then it generates video?” Yes, it generated video, and I managed to get a video of the blog post content created by NotebookLM.

[09:32] But then I stretched it a little bit better, further, and I used Remotion to actually create social media reels, Liz. That was outstanding. And when I say it in tourism, I use Remotion to create content, it used our pictures, you know, our existing pieces of video. It didn’t generate the fake people video. So that’s what’s really important.

[09:57] We’re at the stage now that AI can go and grab your stuff from your computer that you might not have organised very well, look at all your photos, your videos, rename it, and knows that this is the content of the blog post, and therefore, “I’m going to grab that video, and that’s the piece of information that’s going to really cut it for TikTok as opposed to Instagram.” So yes, that’s where we’re at. Obviously, I’ve got a tech background. I love spending a bit of time playing with these tools, but I have not spent any money apart from our monthly Claude subscription on any of that. So it is totally possible to be done.

[10:35] Liz Ward: I think this is another good… this is a great place right here to say something I’m really proud that you have developed, that I’ve enjoyed showing to tourism operators, is an AI app, or it’s an… it’s being an agent, which is Robin. So Batman’s sidekick Robin, who can create… take an idea that a tourism operator has and create a script and a hook for a social media reel. So that is a fantastic little tool. That is a great example of AI as an agent that anyone can use, and you’ve developed it for them. And anybody can jump on and have a look at that, getting that from the tourismtribe.ai website, to have a play with that.

[11:28] And so, showing operators how they can just drop their ideas or save their ideas, and it’s already pre-briefed with all of their context, their marketing, target markets, their brand, their brand voice, all of that, and create that. So that’s another example of using AI smartly on your P-plates to be able to really get real value from it.

[11:52] And I’d like to share my own personal journey, because I produce a lot of slide presentations for training, and pitch documents and proposals for responding to opportunities to win Tourism Tribe work in programs. I set myself a personal goal of moving off… I didn’t call it moving off my L-plates in January, but that’s what I was doing, moving out of my comfort zone of ChatGPT. And I set a goal of dumping everything I knew about how I used to produce those types of outputs, like presentations and proposal documents, and I started exploring. And so, where I land, and this is what I mean about you’ve got to explore, you’ve got to try and work out what’s going to work for you, and for me, it is NotebookLM. So, notebooklm.google.com. If you’ve got a Gmail account or a Google account, you can get in there and have a play.

[12:56] Now, this is an amazing tool to be able to curate all of your sources, all your pre-existing knowledge that can be sitting in other documents or web pages or other sources external, ring-fence all of that, and let it… get it to produce for you what the output is that you need. Whether that is a slide deck presentation, it could be an infographic, it could be a document, could be a video, and you get an amazing output. You can instruct it on your brand and upload your style guide so that you get a look and feel. So that was an amazing change for me to realise the potential, but it actually hurt my brain to drop the way I used to do it.

[13:42] Fabienne Wintle: I know the feeling.

[13:43] Liz Ward: Which was to go in and copy a previous document and then edit it. And I might have used ChatGPT in the process, but that transition was amazing. And then the other thing that has been really helpful to me, which can be useful for tourism operators prepping for trade shows, or they might be doing a presentation for whatever reason, is a tool called gamma.app, to be able to produce beautiful documents and presentations. So, from scratch again. So it’s another one that I’ve found really useful to be able to do something extremely polished-looking, and very, very quickly. So, just thought I’d share that as a couple of examples.

[14:27] But what I’d like to do now is move on to what does it look like on a full open licence in a practical way, and some… and examples. But maybe just start, Fab, by telling us about your journey since late last year in being able to understand the capabilities of agents and the software development capability of the tools that you’re using. And to just remind people that you have a web developer background, a degree in IT as well as tourism. You’ve had your own web development agency, so you have all of that knowledge and experience that gives you a wonderful base to draw on when you’re evaluating these tools and going through these exercises. But I think if you can give us a snapshot of what that has looked like, and then we’ll talk about a really good real-life example of solving a problem.

[15:23] Fabienne Wintle: Well, the first thing people need to realise is AI will code things for you. So the need for developers is almost redundant. And I think 80% of the code used by Anthropic, which is the company that owns Claude, is AI generated. So the skill set… thank goodness, I never really liked coding. I was never interested in learning all these extra languages because I wouldn’t remember them very well. The skill set required now is not to write the languages, is to coordinate, orchestrate, very similar to a software development agency’s boss that needs to really understand, plan what needs to be done, and brief the coders, which is now the coding tool or the coding brand, if you’d like, that we use. So that was really good for me, because there was so many things I wanted to do for us at Tourism Tribe that I could never do because I didn’t have that knowledge of the code anymore. That was long gone.

[16:34] So I just, like you, Liz, I just put myself… I grinded through the paces that I needed to do, because you can see an example, but for me, I really have to practise to understand it. And there’s all these influencers on TikTok saying, “Oh, you can do that and that.” But a lot of them are influencers that just repeat what another one has said, but no one actually gives you the how-to. So once you’ve worked out the how-to, things get much easier.

[17:05] And what really made it click for me to be able to create that, and literally now I’ve got, while we’re talking, I’ve got AI agents doing my work in lots of different areas. So I, you know, got down to that point, but that was really late last week. They’re not running on my computer, they’re running on another machine, and it’s all happening, and I’ve got full control, and they ping me if they need me on Telegram, so I can see what’s happening. So this is where I’m at now.

[17:33] But my… what really helped me understand is when, you know, these AI agents is… well, it’s not a big box. Like, you actually have to work out the process, and an AI agent is actually like an employee in the company, and they have skills. So when you break down AI agents, the black box, what is it? And you visualise, actually, I need lots of people to help me with these jobs. I don’t want them to be me, because that’s the whole point. I want to be able to do other things, so I’m going to manage them, but to manage them, I need to understand what skills they need to have. So I need to understand that an agent is like an employee. If the employee is not qualified, it doesn’t have skills, the output is not going to be good.

[18:23] So when I started visualising these AI agents, well, actually, I’m building a software development company, or I’m building a web development company. I need the web developer, I need the back-end web developer, I need the front-end web developer, I need the accessibility checker, and these are the skills that they need to have. Everything started working beautifully. But when I talk to, you know, I put things on social media, and people want to pick my brain all the time, and I’m like, “So, have you got the agent and the skills?” They’re like, “Oh, no, no, I’ll just ask Claude.” I’m like, “You’re actually asking the receptionist to give you advice on whatever you’re asking.” So that’s not the way it works. So once I explain to them that they’re using the receptionist, and they’re not giving it any context, well, the output’s not going to be very good. It’s not going to be consistent. They get it.

[19:12] So the concept of you need a team of agents, that means you need a team, a staff team that are skilled and they have skills, that made it work. And what’s an AI agent and what’s a skill? You know what it is, Liz? It’s just one document that explains what an agent is, following how Claude Code writes it. And a skill is another document. Like, they’re called… they’re like dot… documents, but they’re called MD files. So once you get that and you’ve got all the tools set up on your computer, which might take people the longest, really, you get great outcomes.

[19:50] Liz Ward: So that’s an amazing learning and experience you’ve gone through, and now you’re turning out some incredible work. So I saw the website that was a rebuild for an operator just this week, which was incredible. And if you compare that to, like, how long did that take you compared to what it would have taken last year?

[20:14] Fabienne Wintle: It would take me a good week to do a rebuild like this. It took my agent maybe max 30 minutes. I was doing something else when the agent was working.

[20:25] Liz Ward: But you’d obviously briefed it. It had very good instructions.

[20:28] Fabienne Wintle: Oh, the briefing, yes, they are IP, you know, that’s the thing. That’s what we do. So that’s why it’s not hard for us. Once we understand the process and how to coordinate, it’s not hard because we’ve already got all the IP of all the steps. A lot of it is in my head, so I had to extract it and make sure it’s up to date. And obviously I get Claude to challenge me because I don’t know everything, right? Why not? Let’s use the top model. That’s another thing people don’t understand is you choose ChatGPT, Gemini or Claude, but they’ve got different models within themselves for different things. So you got to use the right model. But once you’ve sat down, you know, many nights and many early mornings, and you figure that out, you feel like, oh, okay, yep, I can do this, you know. But you’ve got to work through the process.

[21:16] Liz Ward: Amazing. Amazing. So what I’d like to finish on is just this week I met with a wonderful tour operator, 4.9 stars, hundreds and hundreds of reviews, in regional Australia, and it is a tour company that takes out groups exploring via bike. And he has a wonderful problem to solve, which you are going to be working on for him, which I think that a lot of tourism operators can relate to. So he works with inbound tour operators for his inter… some of his international market, and he has a significant international market. And those inbound tour operators, and this is typical within that sector of travel trade, a lot of them do not have systems. A lot of them rely on their personal relationships, and email is their communication tool and booking tool of choice.

[22:14] And so he receives multiple times a day requests to actually confirm great business of a number of people on a certain day to do that tour. Now, his problem is that he has to manually go into the front end of his website to do the booking as if he was an agent. And those agents have a coupon code that goes in so that it’ll… because he has to generate the invoice manually out of Xero for these types of bookings.

[22:47] He then has to go into his res system, which is Rezdy, and trigger something else. He’s got to put in some other data, which triggers notification to the suppliers who are on the itinerary, who he will be calling into or booking for them. And then he has to generate a confirmation notification slash email back to the inbound tour operator, and then he has to generate the invoice in Xero. So all of this is done manually. Xero takes care of pushing the invoice out.

[23:25] He… because this fellow is… this fellow, he knows that he doesn’t know stuff, but he knows what the capability and potential is of AI to solve this problem, because every one of these takes a minimum of half an hour of manual work, and he is a small and growing business that doesn’t want to keep putting a half hour every time he gets one of these and eroding the potential profit and his resources on this kind of mundane manual work. So my question to you, Fab, is where would you start with a problem like that, thinking about AI as a solution, and where do your agents play a role here?

[24:04] Fabienne Wintle: So what works really well, and I mean, obviously you’ve described it. So where would I start? I would have done the recording of this very podcast in his client folder on my computer, right? Because that’s how I do it. Let’s take every ounce of information that we’ve got. Then I would obviously jump on a meeting with him, which I would record, transcribe. Again, that becomes like IP that is going to be critical for Claude Opus 4.6, the model that I would use, and I would challenge Claude as, how would you do that? But obviously, talking about this and being very, very familiar with these tools, I’ve got my idea.

[24:41] Then what I would do is I would send my content audit agent through all his documents, all his website, all the emails, like samples of emails that he may have received, all the whole process, even pictures of them. It doesn’t matter. Like, whatever format, it can encode it. And then create what I call the official documentation about the business. And the official documentation is their pitch, what they do, their target market, etc., because if I was to brief a senior automation manager, you know, pre-AI, this would be the person, or brief myself, I would have to understand the business and the context. And that’s where people make a lot of mistakes, whether they are in ChatGPT and they ask the prompt for a blog post, but the thing sounds a lot like AI, not like… missing opportunities on their business because they haven’t given it the context.

[25:32] And the context window in the memory of AI is getting bigger and bigger, but you need to create that documentation. I visualise it like the sales… door-to-door salesperson going with their suitcase, you know, every time they’re meeting someone. It’s, “Well, here’s what we do.” That’s the context. So, generate all this, and then the operator’s got those 25 files available to them to do anything, write blog posts, it’s there. It’s about that. So, create that context. That’s the base for the agents.

[26:03] And then, well, I would… to me, it doesn’t sound like all of it would have to be AI. There’s a lot of tools that already might have integration that are possible to use, because every time you use AI, it counts as a credit, or they call it a token. So you need a membership to the different systems, or you need to pay API costs. So the least possible API… AI that you need to do to reformat the information, the better, because there’s obviously less chances of going wrong, and it’s cheaper because you might not need the tokens. But you might use AI at the beginning, and as I…

[26:44] Liz Ward: Oh, for the planning?

[26:45] Fabienne Wintle: Absolutely.

[26:45] Liz Ward: So you’re using AI, and I think what’s coming through here is the importance of that briefing. It’s like if you compare it to, as you say, the old days of business analysis, and having a business analyst, and they would have gone through that process to understand that, and there would have been spreadsheets and documents and every conversation recorded. And it’s using AI for that really heavy lifting, detailed work to be able to get a good understanding. But I’d imagine that you could use an AI agent as a checkpoint, or checking that all of this worked. If you’re using Zapier or other data parsing tools or automation tools, you could actually use an AI agent, couldn’t you, to sort of look over the process and ensure that it’s happening.

[27:32] Fabienne Wintle: And you might not even need Zapier. The systems may integrate directly. It might be just a matter… So it’ll be a mix of different systems, of way of doing things. But yeah, I would have to break it down, even the agent, you know, hand it off to someone else to do. It’s… Remember, AI has replicated the human brain. So it had to have a model, which is us. So you… Once you realise, “Oh, yeah, well, I’ve got to break it down into tasks so it can pick it up.” And then it needs to understand that this one needs to happen after this one. But yeah, that briefing is critical.

[28:07] Where it keeps the cost down now with AI is us knowing how to get that briefing automatically, you know? And we can spend so much more time with clients on things that AI still can’t do, or get ideas, or, you know, look at their business and feed that to their AI. But that planning piece is where… and you can watch as many TikToks as you want and you look at success stories, that’s where things go wrong, if you don’t do it properly.

[28:36] Liz Ward: I think that’s great. Thank you for sharing that, and consider yourself well briefed on that problem that you’re going to solve for that tourism operator. So to wrap this up, I think if we say to people, please don’t feel like if you’re not developing agents, you’re not there with AI. You have to progress through, make the most of the tool you’re using now, work out how to get the most out of it, but then stretch yourself. The next time you go to do a task, ask yourself, how could I use an AI tool to do this for me? And as painful as that is to give up the old habits, take the time to explore, or have a chat with us, and we’re happy to guide you.

[29:20] We’ve got lots of examples we can share, lots of… happy to give you advice. We’ve got a brand new back pocket tool in beta, ready to come to you, which is my.tourismtribe.com, which is going to put us in your back pocket so that you can interact with us. We’ll share our free tools through there, so that will make it much easier for you to access the knowledge of our community and our team and our AI team, our AI tools to be able to help you. So jump on to my.tourismtribe.com. It’s in beta at the moment, and we’d love to know what you think. And go on the journey, and we’re here to support you along the way.

[30:09] Fabienne Wintle: And one tip I’ll say, and I’ll say that to our team all the time, is if they don’t know how to do it, I say, “Have you asked it?” Because that’s the best way. But when you ask it, don’t say, “Oh, I don’t know how to do that. How do you do it?” Again, give it just that extra little sentence of context, so you get the best answer. And if you’re not one who likes typing, or types as fast as us, Liz, press that little microphone button on whatever AI you’re in, so you can dictate to it. Unbelievable how many people don’t know about that one.

[30:43] Liz Ward: I know. I know. And I love that because you are so much more verbose and expressive, and you use different words when you talk to explain something than when you write. So it’s definitely worthwhile trying that. Great. That’s a wrap. Look forward to speaking with everybody again soon.

[31:04] Fabienne Wintle: See you. Bye.