AI-powered workflows for marketers

Link to the slides / Hyper-personalised emails JSON file (for make.com) / YouTube to Blog JSON (for make.com)

‍

[00:00:00] Maryrose: It or you could be. You could be, uh, working away in, uh, in a company, you could be a consultant, and you might get an email from a client asking you to knock out five blog posts this month, and you're sitting down, you're working out how long it's going to take you, so you might think about, you know, you're going to have to, you know, research a blog post, you're going to have to write the blog post, get it up on their site, there's all the social media captions to do.

[00:00:24] Maryrose: Bit of to and from with the client. So if you reasonably assume five hours per post, that's about a 25 hour piece of work. And we're assuming somebody is charging out at 75 euros per hour. Very good value. If you can find that person, snip them up now. So you would look at charging the client around 18, 1875.

[00:00:44] Maryrose: Okay, that was before. Now, after. What happens next? Well, in this new world, you can reduce, or that person, Claire, can reduce her hours from 25 to 6. She's still going to charge the same, if she likes. Or, she can re divert the time she saved into doing something else instead. And, we know, there's probably going to be a better output.

[00:01:07] Maryrose: She's probably going to, you know, Produce something that's better than what you would have done had she just started doing the five blog posts on her own. So that's what we're looking at today. How to help you do that. So my name is Mary Rose Lyons. I'm delighted to be here. I'm delighted to be a member of this community.

[00:01:23] Maryrose: Uh, I love Ben's Bites. I love everything you do Ben and team. You're great. I'm the founder of the AI Institute. We're based in Ireland. We do a lot of work in Ireland, the UK and Europe, and I would describe myself as a digital marketing OG. So that means that I've been around for a very long time doing digital marketing.

[00:01:41] Maryrose: In fact, I remember when Google IPO'd and I'm very excited to be here. To be working in the space, particularly in the last couple of years. It's an exciting time to be a marketer. There's so many great things that we can do to change up the way things have always been done. So I am a marketer, and I'm [00:02:00] going to introduce you now in a moment to my marketing soul sister from South Africa, Claire Dupre.

[00:02:05] Maryrose: So remembering that we are marketers when it comes to question time, because I watch a lot of these workshops and I know there's a lot of developers Who attend these calls. So go easy on us, please, on the questions because, because we're marketers. Over to you, Claire.

[00:02:22] Claire: Thank you for that, um, Mary Rose. Um, I just want to check, has everyone seen the presenter view, or are you seeing the full slides?

[00:02:30] Ben: I think we see the presenter view so we can see the next slide is.

[00:02:34] Maryrose: Oh dear. I went there. We fixed that. Thanks. Good pickup.

[00:02:38] Claire: So while Mary Rose is changing that, I'll do my quick intro. Is that better? Yes, that's right. Thanks, Mary Rose. Um, so yeah, great to be here amongst members of what is one of my most favorites, um, AI communities.

[00:02:54] Claire: And in fact, Ben, this was the first AI community that I bothered to fork out cash to join. So kudos to you. Um, and I think what really attracted me to joining this community was the focus on. Practical application off of a I. Um, And so that's why I'm really excited to be here, putting my hand up with Mary Rose this evening to talk to the practical use cases that I've been, um, building out automations around in my work as a digital marketer and, um, a I'm marketing consultant.

[00:03:25] Claire: Um, I know you're not here for the intros, and I tend to be boring, so I'll keep it brief. Mary Rose already introduced me. I'm based in Cape Town, South Africa. I work in ed tech, uh, predominantly in brand and product marketing. And for the past year, I've really been diving into understanding how does AI drive business value?

[00:03:45] Claire: Um, I'm really like allergic to the shiny objects and hype of AI, and I think Figuring out how we can see real returns from AI is, is really what I'm interested in, and I think that's really our intention for today is to demonstrate [00:04:00] some of the real returns that we've been seeing from AI automations in, um, the marketing work that we've been doing.

[00:04:07] Claire: Um, And then sharing those learnings with you in this format. So we're gonna, we're gonna share two use cases for no-code AI marketing automations together. Um, and we're gonna demo how we're using make.com to do those marketing automations. And then we're gonna pull together, we're gonna give everybody an opportunity to join us in actually building this in make, um, or alternatively just watching how we do it.

[00:04:31] Claire: And, um, hopefully you'll, you'll walk away with, with some better insights and AI automation.

[00:04:37] Maryrose: And hopefully the demos will work. You can always rely on a little bit of technology to go wrong. Just to mention actually, um, we have, Claire and I have actually put our collective knowledge into a course which is for marketers.

[00:04:50] Maryrose: Any marketer who's already using AI and wants to transform the way they work by actually bringing all of it together into AI automations, like what we're going to show today. We have a course, which is made up of learning sessions and then one to one workshops where we really get under the hood and help you get your stuff working.

[00:05:08] Maryrose: There is a discount code, always give away a discount code. And the other really great thing is that the blueprints of what we make today. So that's, in other words, the, the JSON file, we're going to be, um, sharing them back out, uh, through Ben and Amy, so that you can take our blueprints, import them into your iteration of make.

[00:05:27] Maryrose: com and, uh, just take the work and just roll with it. So that's what we're going to be looking at. Just to be very clear on what we mean by, uh, AI automations. Anyone who is a marketer will be familiar with just general marketing automations. They've been around a very long time. You know, examples might be things like when somebody signs up for your email newsletter, you know, best practices to send an automation out about six hours later.

[00:05:53] Maryrose: You know, talking to your brand experience and what they can expect from you in the email and other automations would be [00:06:00] familiar with would be things like auto posting across, you know, social channels. I don't agree with it, but there you go. Um, but when we're coming in to say. Um, AI enhanced automations, that might be where you've started playing around with using like perplexity to summarize something.

[00:06:17] Maryrose: You might go over into Claude to get it right. A really nice caption for LinkedIn. You might hop into, you know, Gemini to get to do something for, um, Instagram, for example. So that's great. And that's, that's a really good place to be. But the next level up is where you're actually going to automate that entire process.

[00:06:36] Maryrose: So this is what we're referring to when we're talking about AI marketing automations. You're taking a particular piece of work that you as a marketer would do. You're already augmenting it with AI, and then you're going to use something to bring it together. So that's where. Make. com comes into it, so Claire and I bonded over a shared love of make.

[00:06:58] Maryrose: com make. com is an automation platform that is especially for no code automations, meaning marketers like ourselves can actually do it without needing to be able to code. I can't code, but I can do these things on make. com. The reason we love make. com is because it's actually really easy to use. The UX is really good.

[00:07:20] Maryrose: Their support is really good. And the pricing is rather brilliant. It's a lot cheaper than longer established tools like say Zapier, for example, and on the free account, it's a generous free account, which I always love. So you can do two scenarios. That's like two automations. We can do two scenarios a month with about a thousand operations on it.

[00:07:42] Maryrose: And then when you go into the paid version, it's still quite cheap. It's like 10. 59 if you're paying monthly and that's the dollars. Or if you go annually, it's nine a month. Now, I know that Lindy and Gumloop are also parts of the Benz bike community. And I've actually watched back on all of the sessions that they've run.

[00:07:59] Maryrose: They're [00:08:00] also really good tools. So if you are looking into this and you're at the beginning of your journey, And I'd suggest you just have a play with either Lindy Gum Loop or Make, and decide which is the one for you. It's very important though that you make sure that the tool that you do choose has the right amount and the right APIs that you actually need.

[00:08:20] Maryrose: So over to you, Claire, on APIs.

[00:08:23] Claire: So we just included this, um, word about APIs with the view that there may be people on this call who, like myself, about a year ago before I You know, came and found out about the world of AI automation. I didn't even know what an API was. And I was actually terrified of this.

[00:08:39] Claire: I thought like the technical side of this is not for me. Um, I'm never going to do touch stuff like this. And turns out I was afraid of nothing. Um, and API is actually really simple. So by the off chance that there is someone in the school that is feeling like that, um, you know, in the simplest form, an API is how applications communicate with each other.

[00:09:01] Claire: And it's how these. Automation integrators work, how they connect the software that talks to each other and connect, um, the tools that are going to work without you intervening in the automation. Um, I think one of the most important things about APIs is it's one of the reasons that I landed up with make.

[00:09:17] Claire: com as my integrator of choice. And that was because I had a desire to take my AI note taker at one point in time. And have it automate meeting notes into Slack in a very specific format. And in doing my research across all of the different automation platforms, I couldn't find. Any of them that had an API that was going to connect with my AI note taker until I found that make.

[00:09:42] Claire: com had a API for fireflies. And that landed me in an air and a fireflies account and I built an automation and I was just like, yes, this is exactly what I'm looking for. So the API is actually quite a big. Decision factor for which tool you're choosing. Um, and you can go and see the whole list of API [00:10:00] APIs that these integrators have, most of them list them on their sites and they're updating them all the time.

[00:10:05] Claire: Um, I know Mary Rose and I, Mary Rose, what was it about Canva? When we started working on make, we were so frustrated that there was no Canva API and recently, yeah, recently I think they launched their API. And now we've been able to play a little bit with Canvas API. So. Changes all the time, right?

[00:10:24] Maryrose: Every, no, but like we were kind of saying, look, every marketer is using, well, most marketers are well used to using Canva.

[00:10:30] Maryrose: And the fact that Canva were blocking all but their enterprise customers, you know, the ring us for a price customers. Yeah. Um, so they're blocking all but them from using their API. And we were very angry about this. Um, we were taught tossing and given out about it. And then one particular day we just opened up make and there you go.

[00:10:48] Maryrose: Canvas in there. So it's really important to keep on looking and see what, seeing what they're due to refresh the different ones that they have on. So, yeah. So moving on to, uh, Claire is going to do a demo. She's going first. Go on Claire. Woo.

[00:11:10] Maryrose: Oh, I'll stop sharing. And then we can hop over into your screen.

[00:11:13] Claire: Okay. Can you see my screen? Which screen are you seeing?

[00:11:18] Maryrose: I'm going to stop sharing. Oh, what a tidy desktop you have. I'm really glad we're not looking at my desktop.

[00:11:27] Claire: Hi. Come on. There we go. Yeah, there we go.

[00:11:30] Maryrose: Great.

[00:11:31] Claire: Okay, so you can see my screen good.

[00:11:34] Claire: All right, so i'm just opening up make. com and the Automation that I wanted to showcase this evening was one that essentially creates Hyper personalized email outreach to new leads that come in. Um, Let's call it from your website. And, um, I want to just pause there and talk to you, like, you know, the narrative of business impact, why this was an important automation for me to design and build in the work [00:12:00] that I did was when my team and I decided to start doing hyper personalized email outreach, we were finding.

[00:12:08] Claire: Results like increases of 20 percent open rates, increases of 40 percent engagement rates, response rates, conversions to customers. It was insane. Like the value we could get from customizing emails and being hyper personalized was incredible, but we had a pain point. We couldn't do it at scale and being a small startup team.

[00:12:28] Claire: We, um, we're taking like three days, three of us to sit and research, write, customize, edit, and send emails to get the results that we wanted off the hyper personalized emails. And we simply just didn't have time like that. That's that's time is money and startups specifically. So I thought, well, could AI help?

[00:12:46] Claire: And that's what led me to building this very first, um, AI automation that I had in Meg. So this isn't the exact one that I run in my organization, because that's connected to HubSpot and contains sensitive customer data, but it is a similar one. Um, so I'm going to show you quickly how this works.

[00:13:03] Claire: Essentially, it starts the trigger of this automation is it's connected to a Google sheets where all the lead information comes in, and this could be anything. It could be a table. It could be your HubSpot CRM directly. But what happens is as soon as a new lead comes into the sheets, um, it will fire off this automation.

[00:13:22] Claire: So let me just pull this across so you can see this as well. So this is what that sheet looks like. So. I'm collecting, call this, I'll leave capture form field from our website. And I'm collecting a first name, last name. What's your goal? What's your title? What's your company? And what's your email address?

[00:13:39] Claire: And this is the data that we're going to pull into the scenario. So in a scenario, then what happens? I'm going to run it once so we can just watch it in action. So it gets a new lead from the Google sheet. We're then going over to Perplexity and saying, Hey, can you go onto the internet and do some research for us on this specific person, their job title, their [00:14:00] goal, and find some interesting insights that we could use to write a compelling and persuasive personalized email for them.

[00:14:06] Claire: We're then going to hire Claude as our content writer, who's going to write the email for us. And then I've also asked Claude to come up with a compelling subject line with a prompter that looks at best practice hooks and open rate conversions. And then we're going to spit that out as a draft in Gmail.

[00:14:23] Claire: Definitely not confident enough to let AI just run an email without me checking it in draft first. Um, I don't know if anyone else is wilder than I am. Um, but that's essentially what it does. So, and I'll show you what, what that actually looks like when we come over to, um, the drafts in my inbox. So we just run that automation ourselves and we pull this across.

[00:14:45] Claire: And there you can see the emails popped up in my inbox, which is unlock your data analytics potential quick call and I'll open it up. There we go. So perfectly already in my inbox. Uh, great to go. I can edit it, customize it, et cetera. Uh, it saves me so much time. So. Coming back, I did see there were some notes in the chat, were there any questions I missed?

[00:15:09] Claire: No, always draft 100 percent Ben. Um, cool. So coming back to the actual automation, let's build one from scratch together so that you could do this yourself. If you would like to know how to, um, like Mary Rose did say, though, we will be showing the blueprints. So lucky for you, you get to just import this and connect your APIs and run it yourself.

[00:15:28] Claire: Hopefully it works. And I'm going to open up a new make scenario over here. Let's get my screen sorted quick and pull this one across. Okay, believe it or not, I don't tend to use Zoom that often. Um, sorry, this Zoom setup is bugging my screens. Okay, so we are inside make. com and I'm going to create a new scenario.

[00:15:56] Claire: And first thing, best practice is just to call it what it is, a [00:16:00] hyper personalized email outreach to leads. Now our trigger event is going to be to watch for a new sheet, a new row in a Google Sheet. So we've got our Google Sheets module that we're going to add. We're going to add, watch new rows. And now I won't be setting up my Google connection because it is already connected, but if you're new to make, you would just have to do a simple connector between your Google account and your make.

[00:16:28] Claire: com account. Pretty sure. Um, there's some easy tutorials on this. If you need to, otherwise I'm on Ben's Slack community. Please reach out. If you're having any trouble setting anything up, Mary Rose and hi, are here to help. It's now already pulled through my Google drive. So I'm going to choose that. Form that Excel spreadsheet or spreadsheet that I demoed earlier, which is called a BD demo.

[00:16:52] Claire: And I'm going to connect that here and it's going to pull information from that sheet into make. Sheet name is the first one. And I'm going to set it to limit of one because I'm only pulling through one lead at a time for the purposes of this demo. But if you were doing this at scale, you could have like it writing 50, running this 50 times and spitting out 50 emails quite fast depending on how many leads, the volume of leads are coming through.

[00:17:14] Claire: Okay, so that is our first module. And I'm just going to choose manually where to start for the purposes of this. And I'm going to start right at the top of my leaves and Hey, there we go. So now we've got our very first module. That is our trigger event. I'm also going to change how often this runs and I'm going to make it run on demand, but you could change this to run at regular interview intervals, every day, day to the week, there's a lot of customization that you could set this up and obviously it's going to cost you every time it runs, but in this instance, for the purposes of the demo, I'm just going to say when I hit run, please run.

[00:17:51] Claire: Okay. So we've got our first module. Now, what we want to do is we're actually wanting to go and do our research in perplexity. So [00:18:00] interesting thing about perplexities module in make is that it's the only, um, AI module that can run live. Search on the internet. So chat GPT can't do that. You actually have to download an HTML file and spit it into, into chat GPT.

[00:18:15] Claire: If you're working with live data from, from the internet, which is why perplexity is quite a nice one to use. Um, I'm just going to set my max tokens to like one, five. You could just set this to the maximum, which is 4, 000 or. Limited if you would like to, and temperature, which is where we want to control the variability and the creativity of the output.

[00:18:32] Claire: I'm going to set to one because I'm actually looking for stats and factual information in the research that perplexity is going to go away and do. And then we're going to add our prompt, which we add in here. And firstly, I'm actually just going to copy it across in the interest of time. So this is the prompt, and this will all be here in the blueprint for you, but let me show you what the prompt is.

[00:18:55] Claire: So I want you to research and identify five interesting opportunities in the field of data that align with the career goals of, and now I've mapped it to the information in the Google Sheets. So, So if you click here, all these green little variables come up and you can pick the information from the Google sheet that you're wanting to use in your prompt.

[00:19:14] Claire: In this case, I've got job title and goal from that initial lead capture form, uh, information. And I've given Plexi some direction as to what I wanted to go and do, how it should approach finding the facts and the stats, and then guidance on how I wanted to present the facts back to me and say, okay, And what am I missing here?

[00:19:33] Claire: Our role. Okay. So we're just going to put user as the role. Okay. So I'm just going to run this once quickly, just to show you the input. That comes out of this and you'll see, if we look, if we check the actual run of it and the operation, it's pulled through the first lead. You can see my name comes through my goal, my job title, the company I work for, my email address.

[00:19:52] Claire: So that's pulled through and now perplexity has gone away and done its thing. And we can actually check. The work that it's doing and it [00:20:00] said here are five interesting and practical opportunities in the field of data that align with career goals of the marketing director looking to transition into a data driven role so we can just check that it's actually done that work for us as we go.

[00:20:11] Claire: So now we've got our information that we're wanting to use in writing a hyper personalized email. So the next step is to bring up Claude. I prefer Claude. I just like the writing style of Claude. And again, I'm just going to use 1500 tokens. And we're going to put in our prompt. Um, once again, I'm going to copy it to save time.

[00:20:36] Claire: Um, here. And what we're asking Claude to do for us in this instance is You are, let me just align this so you can see, okay, you are an expert email bot specializing in creating high quality, highly hyper personalized outreach emails using a provided template. Um, you've got key information about a prospective lead, their job role, their goals, and you're going to help create this relevant email.

[00:21:06] Claire: Um, I've given it the fields that I wanted to use from the initial data in module, which is their job title and their goal. And I've also asked it to use the information that came out of complexity right at the bottom. So use this research in the copy that you're going to write. You'll also notice that in this prompt, I've specified the HTML format that I want the email to be written in.

[00:21:30] Claire: So a nice hack for this is that when I create these prompts, I'm actually going into Claude or ChatGPT and asking it to help me write the HTML and then just typing it in here. So I don't know how to code and I don't know HTML. So It really works well to use them as a tutor in App Perspective. Okay, so now we have got our prompt for Claude.

[00:21:52] Claire: And then our final one is to actually do a subject line prompt. So this one I'm just going to say 1000 and I'm going to go through this one quite [00:22:00] quickly. But this one what we're saying is, Hey, can you please also come up with a very compelling subject line? And Split that out for us based on best practice hooks.

[00:22:11] Claire: And in this prompt, you'll see I've got like tons of different hooks, um, that it could draw from, um, and use. So in this case, we're mapping it back to the initial text response from the previous email. Uh, so we're looking at what that Claude module outputted. In the previous module, and we're looking at the summary of the content from, um, perplexity as well, and we're saying, okay, and then finally, we are mapping it to sending our Gmail draft.

[00:22:41] Claire: Okay, create a draft and here we are going to connect. It's connected again to my gmail account. The folder i'm picking is drafts In my inbox or in my email account the recipients i'm going to pull all the way back from that first file uh, which is The way is the email address. There we go subject line.

[00:23:02] Claire: We're pulling from our flawed response Uh, which I think is this one. I can't quite see. And the content is this one. Um, yes, that's right. And we say, okay. And moment of truth in true live demo format. Let's see if this works. I ran on demand. It's going to run once and it's going to do its thing, and I will bring up my inbox, see if it spits it out there.

[00:23:35] Claire: We need a drumroll sound for this part. Dun da da da da da da da da

[00:23:39] Maryrose: da da da da da da da da da

[00:23:40] Claire: da da da da da da. And let's see. Okay, it's still running. It's taking a sweet time. Um, funny how impatient we get. And there we go. Uh, here it is in my inbox. It's just popped up. So unlock your data career.

[00:23:55] Claire: Engineer's path to analytic success. So it's customized the subject line because this person [00:24:00] said they were an engineer. Um, they're looking to move into a data career and it's given us some of that real time content from perplexity. So I know that was like, A speed demo of this, um, automation and because I want to give Mary Rose the floor to do her demo as well.

[00:24:16] Claire: Um, but if there's any questions that anyone has on this automation, happy to chat at the end of the session or by the stack community as well.

[00:24:24] Maryrose: Yay. Well done, Clare. That was brilliant. That was so cool. So, um, let me just share my screen. Can you see the slide deck? I'm just not used to zoom either. Yeah, good.

[00:24:35] Maryrose: Thank you. Okay. So Claire is brilliant in action, and she normally speaks a lot slower than that as well, actually, just so you know. So again, just to mention, if you want to get Claire and I, uh, taking you through how to create all of your own automations, we have this course starting on the 22nd of October, and it's going to be great.

[00:24:53] Maryrose: It's going to be mega. We're going to get people along the journey really quickly. So what I wanted to show you is, um, how we're doing stuff in my business about our new podcast. So we have a new podcast. We've had four episodes drop, as they say. And, um, one of the things I know to be true as a marketer is that You know, you need to be repurposing your content all of the time.

[00:25:17] Maryrose: Um, for example, Google loves when you have a podcast drop on YouTube, for example, and then you carry it over into a long form piece of content for your website. It gives, you know, great, great points for that. But it's the kind of task that I don't really want to sit down and do. And I've never really had the time to kind of, you know, get everything done.

[00:25:35] Maryrose: But now we can. So what I'm going to show you is I'm going to take you through this great demo. So Um, of how to take a piece of video from YouTube to carry it over, create a, um, synopsis and then make a blog post and publish it straight onto your website. Okay. So let's hope that the old, uh, the gods are with me as well.

[00:25:58] Maryrose: Um, I'm just going to [00:26:00] share my actual screen and not the slide deck. Um, where's it gone now? I'm sharing. Okay. This one. Great. Okay, so this is into, into my area. Can you see my, my scenarios? Okay, Grant. So basically this is in to where I am. So just to show you what we're looking at.

[00:26:25] Maryrose: We're going to basically, we're going to go from YouTube. We're going to extract the information. Now there's one little cheat I'm doing just for time. And this is basically going off to an API. That's especially for taking transcript scrapes from YouTube. To actually set this up live would take me about 10 minutes.

[00:26:43] Maryrose: And I thought, I don't want to bore you all with it. So I'm going to start there. Okay. And then we're going to take the content from the transcript. Okay. We're going to get Claude to come in and do his thing, right? A lovely piece. And then we're going to bring it over onto our website, which is built on Webflow, and then we're going to come over and post.

[00:27:01] Maryrose: Okay. So just to show you an example of, um, what this looks like. So basically we had, um, a podcast episode go out this week and, um, with great Julian Barra, great creative. So this is what it created. Know. Pretty good. Um, then we had a podcast episode go out a few weeks back or last week, sorry. Barry Scannell then, and I'm actually leaving these in to show you.

[00:27:26] Maryrose: We had, um, Claire on the podcast and we, another great lady called Lolly, and this is where I hadn't actually fine tuned the automation. I just set it up. So we got like about eight to 10. Of Lolly and Claire, because I kept pressing buttons and I didn't know the thing was running in the background. But I left them up anyway because they had like really good things to say and to share.

[00:27:48] Maryrose: Um, so if you want to catch them, you can. Okay, so basically, um, we're going to go back over in here. Uh, we're going to go into scenarios. And in true, here's one I prepared earlier. I'm just taking [00:28:00] this little short one here. It's, it's just got the, the only thing I'm cheating on is the API is already set up.

[00:28:05] Maryrose: So as you can see. We have our YouTube channel set up, OK, and we have, um, I'm, I'm asking you to only look for videos that were published after the, I want to take it actually after the, um, the 7th of October, OK, because I don't want to go back into really old ones. I want to, there's one I put up today, actually, which is next week's podcast.

[00:28:27] Maryrose: So I'm actually working while I'm demoing to you. So. Multitasker, um, I'm going to set it to one. Um, as Claire alluded to as well in her demo, one thing I'm really conscious of, I don't know if you've seen this a lot, you know, all these screenshots that are appearing on Twitter or people getting charged like 700 because they left things running well, I'm actually really putting in limits on all of my stuff so that I don't get hit with bills such as that.

[00:28:51] Maryrose: Okay. So to build out this demo live, we are going to add what's called, I'm going to move all of you here. We're going to add an iterator. So the iterator is literally something that's going to come and it's going to have a look at the, um, output from the API, which is I'm going to take the content and then I'm going to hit OK.

[00:29:11] Maryrose: That's just telling me don't end now make us, you know, nice and helpful. Okay, that's an error there, but don't worry, it'll disappear. Then we're going to go into the built in tools. And what I'm doing here now is I'm actually taking a text aggregator. So I want to go back. I wanted to look back on the iterator that I've just created.

[00:29:29] Maryrose: And then I want to tell it what what I want to take in. So in this case, I want to take in obviously the text. Okay, data content. I'm going to hit on to the next one. I'm now going into Claude. Don't we just love Claude? Um, here is Claude. I'm going to, you know, you guys, you're special. I'm going to give you 3.

[00:29:49] Maryrose: 5 sanity. OK, but I am going to maximize out my tokens and I'm going to get it to create a message. I'm going to get it to be a user. I'm going to have some [00:30:00] text and then I'm going to put in my, uh, my fabulous prompt. OK, so I have the prompt over here and it's in the slide deck. So, um, you are all welcome to be treated to the slide deck.

[00:30:15] Maryrose: As well, if you want to, um, okay, so basically I'm going back here and, oh, I didn't say that. Sorry, one, two, oh, and I've copied the, uh, prompt now, and I'm going to put it in correctly. There we go. We're on text. Right. So in my prompt. Boom, what I've told it to do is I've said, you know, I want a well structured HTML and post.

[00:30:41] Maryrose: I said for each topic that is discussed in the podcast, I want you to do a little bit of an expansion on the topic and give some examples or some insights on it. And I've told it to write in British English sort of annoys me. It ignores that a lot of the time. Um, I've given it our own, uh, tone of voice, language, style, and given a few other things about word count and so on.

[00:31:00] Maryrose: I told it to do long form because it understands long form better than it understands word count. So that should have taken the transcript and put it into a gorgeous, uh, post. Now I'm going to bring it over onto our website, which is built on Webflow. I'm not going to lie, it was kind of hard to connect Webflow properly.

[00:31:19] Maryrose: So one of the things I would say to you is that like, don't, don't underestimate the time of just getting your connection set up for the first time. And once you have them set up and you have all your little APIs set up and you've got your credit card feeding out 5 here and 5 there, then you too will be able to fly just like Claire and I are.

[00:31:37] Maryrose: And when you join us for the sessions, that's what that's all about. The workshops are all about getting you set up on your stuff so that you're able to fly too. Okay. So what am I going to do? I'm going to create an item, which is like the language for a writer blog post. What I'm going to do is I'm going to tell us it's my site.

[00:31:54] Maryrose: Yeah. Um, it's not archived. It's from, it goes into insights cause we call our blog [00:32:00] insights. Um, and it is basically going to be, so the name is going to be, I'm going to take the title from the original YouTube, and then I'm going to go down into the URL and I'm going to give it, um, the URL. From the content, no, not from the content right there, from the thumbnail, that one here.

[00:32:24] Maryrose: So down here, I'm going to take the URL there. Actually that's for the image and then finally I'm going to tell it to take the text that lovely, uh, Claude has produced for us that is here and I'm going to hit OK. Oh, I should have put in my author name actually. So what we're doing so that we know, uh, Where the post came from because around the Claire and Lolly time I was sending messages to my team saying who's posting all these things about Claire and they're all coming back saying not me no it wasn't me and then we realized actually it was me and my automations so basically I'm just going to put in something like BB here and I'm going to hit okay so now we have the post in Webflow Next thing to do is we're going to tell Webflow to actually publish it.

[00:33:12] Maryrose: So we're going to publish an item, not a site. And we have our site, and it's going to go through a little few hoops. AI Institute. And it's in Insights, telling it again. Which one is it? We're going to point back to the last one. Um, I'm going to click on Map, Item ID. And it's this one here, the Item ID that I created last time and click OK.

[00:33:38] Maryrose: So, now I'm going to just have another little look at this. Looks okay to me. Um, like Claire was saying, when you're demoing it's kind of good to run it on demand. But what you can do is you can set this to kind of like look every day or look every six hours or whatever your frequency is like for publishing onto YouTube.

[00:33:56] Maryrose: So now is the moment of truth. I need my drumroll please. I'm going [00:34:00] to run it. I love this feeling when it's doing its thing. Well, it takes a little bit of time to write something great, but that's okay. It's still only a matter of seconds, lads. I don't know about you. Back to us.

[00:34:11] Claire: It's time to think.

[00:34:13] Maryrose: Chloe is a true

[00:34:14] Claire: artist.

[00:34:14] Claire: It's time to think.

[00:34:16] Maryrose: And he's not like aiming to please us as quickly as the others. He kind of takes his time. But I don't know about all of you who are watching, but like, give me a yes or a no in the chat. Have your family noticed a marked increase in your impatience? In the last while, because mine certainly have, and I kind of started out pretty impatient to start out.

[00:34:37] Maryrose: So, okay, so this is saying, look, the scenario run was completed. So the moment of truth, we will go over to insights. I'm going to hit refresh. Ooh, oh, I don't like the one that's taken. I hate that photograph of myself. Oh, collection. I made a mistake. Right there. Let me go back and troubleshoot this.

[00:34:57] Maryrose: Mortifying. Okay. So I've actually gone back here to what's it taken. It has taken collection. Okay. There.

[00:35:07] Claire: You can't do a live demo without one of them not working. Exactly.

[00:35:12] Maryrose: Thank you for that.

[00:35:13] Claire: It's the curse. Sorry.

[00:35:15] Maryrose: Yes, thank you, and it had to have been me. Um, okay, wait there, that's all fine there. Right there.

[00:35:21] Maryrose: It's in the Webflow one. The Webflow one is kind of tricky. Uh, select from the list and introduce insights. Yeah, we've got that, that, that, that. Maybe it's here. Excuse me one sec. I'm going to open. Ah, yes, sorry. I should have taken the actual text. Right, here we go. And then we click on run again. And it's going to do it again.

[00:35:45] Maryrose: It is going to do it. I have every confidence and faith in you. Just while we're doing it again, um, I would like to point out how much I absolutely love Claude for doing this writing. So I've got other, I've got other [00:36:00] automation set up, which, you know, you could build in onto this, which is where when something lands on your website as a blog post, then you can do connections with like say Claude and ChatGPT and Gemini and then get, um, social media posts written out.

[00:36:16] Maryrose: And then you can automate them so that it schedules at a certain time. So you can build these out to be as large or as small as you want. Okay, let me go back here. Bear with me a moment. Insights has a good, hopefully it'll give me a new one.

[00:36:37] Maryrose: Yes, it has small rant. Oh, great. Thanks for that. And yes, I'm applauding myself here. Thank you so much. So nice looking post. Yeah, lovely. So all I'd have to do is go back in and put in some sort of a call to action. Now, what I have done is, um, I just didn't have the time to do it here. It's a good idea.

[00:36:58] Maryrose: After our little episode of things going on our site, you can add on very easily that it sends an email directly to you and notifies you that something has just gone on the site. And that way you can drop what you're doing, go in, have a look, make a teeny tiny edit, and you're done. So does anyone have any questions on that?

[00:37:17] Maryrose: I know it was very, very quick.

[00:37:22] Amie: Yeah, we had a question earlier from someone about roughly like how much it costs to run these kind of automations. Obviously, it will, you know, it depends on volume. But like, if you're just getting started, you know, doing, let's say, Three posts a week or something, any idea of like what that would cost?

[00:37:41] Maryrose: Well, whoever asked that question, thank you, because you know, it has just made me realize we should have actually ended Clare by bringing it back to the cost for either the buyer or the freelancer versus what it used to be. So as I said, you know, you can, you get your, you can have a free, you could do two scenarios on make.

[00:37:58] Maryrose: com for free, but if you're [00:38:00] paying, it's like nine to 10. 10 and then you pay for, um, credits with like, say chat to BT and Claude and so on. Like I, I've like a tenor on and you know, it's just the longest running tenor. Like it's, you're talking scent. Um, what about you, Claire? What's your massive outlay in order to get your business automated?

[00:38:20] Claire: I don't know how to quantify it. Cause there's, there's multiple automations running in our actual organization, but I can tell you that when we pre paid. The API credits on the LLMs and we, we, um, took out a make account. I mean, we're still not out of the 20 that we put up for the API credit. So it's lasting phenomenally long for the amount of operations that it's been running.

[00:38:45] Claire: So I, I haven't figured out how to quantify it specifically, maybe Ben, maybe, you know, but, um, it's cheap. It's really cheap. I'm so impressed.

[00:38:53] Ben: Yeah, I think it's so marginally small that I don't actually even, I haven't got a clue how much any of it costs.

[00:39:00] Maryrose: Yeah. What I love about it, what I love about it is that like, apart from enabling you to do things like this, that we've just shown you, like the one I've shown you is like to do things you always knew you had to do, but probably never always got around to doing, or you had to pay someone else to do, but Clare is showing you is how to be an actual better marketer.

[00:39:22] Maryrose: And this is where I've had a lot more joy from using make. com. As a marketer, I always have loads of ideas for how I can do things better. And then I never really get the time to get around to implementing them. And now I literally just schedule an hour in the morning to sit down and say, okay, I'm going to make this thing.

[00:39:39] Maryrose: And then it's running, maybe do a bit of fine tuning. And then, you know, week by week by week, you're just becoming better than what you've ever been. And I think that's the true purpose of bringing AI into your workflow, isn't it? It's to kind of take away the dross and to make you be a better marketer or to be a better developer or whatever it is that you do.

[00:39:59] Amie: Awesome. [00:40:00]

[00:40:01] Claire: Rebecca. I was going to, I think it was Rebecca's question in the chat actually that I was going to respond to, um, around, yeah, the other common use cases for make, um, I think it's, I mean, there's so many use cases and I can, I can give some examples like back to my current pain points. So what I've been looking at is identifying where can I resolve pain points in our marketing activity using automations?

[00:40:27] Claire: One of the biggest use cases we're working on now is Automating our entire event repurposing strategy. So, you know, you spend a lot of time doing events. Take this one, for example. The amount of value you could extract through a content distribution strategy that's automated off this content just by an automation would be, you know, super powerful.

[00:40:47] Claire: So that's, that's a use case that I'm busy building out quite extensively. Um, yeah, I think that's.

[00:40:54] Maryrose: Yeah, I've got one another one. So for example, when people do a course with us, they get a course completion certificate. So I basically built a lovely automation that, um, all I have to do is drop the names of people and the course and the date into a spreadsheet.

[00:41:09] Maryrose: And then it goes off. This is where the kind of Canva issue came up because Canva wasn't open. So we overcame it by basically putting our Canva design template in as a Google slide. So it basically goes off, populates. The slides, um, publishes them onto our website. So we have unique URLs. Um, and then basically we can send people their course completion certificates that they can add to their LinkedIn profile.

[00:41:34] Maryrose: So I had someone doing that for me every month and it was taking, you know, as the number of course, uh, people was growing, it was taking anywhere between three and five hours a month. So now it actually takes 12 seconds.

[00:41:50] Maryrose: Any others, Claire? What about your Fireflies one?

[00:41:56] Claire: Yeah, I mentioned that one briefly at the beginning. Um, [00:42:00] that was like an operational, uh, enhancement that I wanted to make. I found that, you know, I was in back to back meetings in a day of work, and we'd cover all, we'd maybe talk about the same issue or the same topic across a bunch of different meetings, and I wanted a way to, you know, Kind of summarize one point of topic or one topic of conversation at the end of the day and spit it out in a specific format that I could just send to my team or to my CEO and say, here's the, here's the output of what was discussed today or what we did today.

[00:42:30] Claire: Um, so I built an automation that collect, collected my transcripts throughout the day. And spa it out into a slack message at the end of my day with a full summary of all the different conversations I had a summary of the decisions that were taken and the next action points, and I could copy and paste that and share it with my CEO.

[00:42:49] Claire: And that's just been like a revolutionary operational

[00:42:56] Claire: where really dealt. Um, so that that one's been really impactful for me from a marketing operations perspective.

[00:43:05] Maryrose: Ben, do you have many? I love to get into. Right, go ahead. I was just saying, Ben and Amy, do you have many set up yourselves?

[00:43:13] Ben: We've got one similar to the course completion one you mentioned. So we have, I think it's a four kicks up the trigger as a form.

[00:43:22] Ben: So we have a, we have a quiz that you have to complete to show that you learned the course material. And then once that's completed, so that's tally forms. That is using Zapier, but I am on a X employee plan. Mm-Hmm. So maybe I'm biased there. Um, but it takes the, so if you've completed the form, it pulls through some, some like Cody bits in it where it pulls through the member ID and things from the site, and then it generates an image all in, there's a, there's a tool called HT ML to CSS that actually is a full HTML piece.

[00:43:57] Ben: So you have to like input the, the right pieces [00:44:00] much like you have on that make scenario where. You input the variables in different parts of that HTML, and then it just generates that course completion certificate and then emails them with a link to download it or gives them the actual downloaded file as well.

[00:44:13] Ben: So that's, that's when we have that.

[00:44:16] Maryrose: Yeah.

[00:44:16] Ben: Yeah. It's very, very good for us.

[00:44:18] Maryrose: It's so disrupting SAS. Like, you know, I would have been looking at, as we grow, I would have been looking at like paying for a course completion certificate service. Now I'm not.

[00:44:31] Ben: It took me a while to get the image right, but yeah, it works now.

[00:44:35] Maryrose: Yeah, you can use them for like more nefarious purposes as well. Like, you know, I've seen people do a whole lot of ones on, you know, scraping, um, Google maps, for example, to get email addresses. Um, if that's your thing, you know, you can do that as well. Um, but the, I think like what, what we're kind of sharing in our, in our course, um, is that really you can do anything you want, like the era of just.

[00:45:01] Maryrose: You know, the only thing holding you back now is imagination is here. And, you know, I'd say everyone in the room here has got amazing ideas for what you can do. Um, and you just don't have to be hamstrung anymore by like, I'm not a developer or I don't have a marketing person. So, you know, the era of the kind of one or two person, um, all around unicorn team is upon us.

[00:45:23] Maryrose: I think

[00:45:25] Amie: there's several questions in the chat around prompts. And so I would love to get your insight into your experience with finding the perfect prompt, how to, um, get Claude to write in a similar tone of you and, and basically like your process to, um, get comfortable enough to give, you know, the prompt to Claude and make.

[00:45:47] Maryrose: Yeah, well, I'll answer first of all, the one on how to get thought to write in a tone that's similar to you. Um, we, I suppose a bit of background there would be that we, we actually have built out some really nice [00:46:00] GPTs on chat GPT. Um, we have, I think every marketer needs at least five to 10 at this point.

[00:46:07] Maryrose: Um, so we have one called voice of us and what we've done is we have, um, we've, we've, we've asked. Chat GBT for, um, it's taken what our tone is based on our website, but we also feed it with links and copy that we like. So not all of it has been written by us. We might find an article written by a really clever journalist and we go, really like that tone.

[00:46:32] Maryrose: So we feed, we continually update this and feed it with the style that we like. And it's got really sharp notes. It's really good. Um, we've trained it on, you know, what style for say LinkedIn posts versus what style for say long form. So all of our team knows to use that. So for example, if I'm actually getting Claude to write something, um, I will actually get voice of us to give me a description of how to write it for Claude.

[00:46:57] Maryrose: And then I bring that prompt over to Claude and I get Claude to write it. I'm a, I prefer Claude for, Writing style in that way, but I prefer chat GPT for the more sort of bespoke settings that you can put in place.

[00:47:12] Claire: Yeah, I want to add to that that in the automation that we are using for these hyper personalized emails in the organization that I'm in, we actually are using a chat GPT assistant as part of the automation that Is trained on the brand tone and voice.

[00:47:30] Claire: So it's a slightly more complex setup that it connects to a dedicated assistant that's been built in the backend of, of chat GPT. Um, and then only going into Claude. So, but I didn't want to demo multitude of other, other names in one automation for today, but that's one way we get, we circumvent that.

[00:47:53] Amie: And then how are you guys, um, measuring, like, impact of these automations? Have you gotten there yet, uh, versus if you were [00:48:00] to do this on your own, um, are you tracking, like, opens and clicks based on, you know, certain prompts or anything like that?

[00:48:09] Maryrose: Um, I'll answer that first by saying that, uh, so we, we do in the AI Institute, like, we do training courses.

[00:48:15] Maryrose: That's the kind of the B2C side, but on the B2B side, we actually go into companies and we go, okay, Let's look at your processes. Let's map them out and break down the tasks and then apply what we know about AI and automation to them. So there is a, one of the very first companies that we worked with, um, is a company called Catalyst Media.

[00:48:35] Maryrose: Um, they are the company behind the Dublin Tech Summit, which is a fabulous event that takes place in Dublin every year in May. Definitely recommend it. Um, so what we did, we worked really closely with, they have a marketing team, they have an editorial team, and they have a sales team. So we worked closely with all three.

[00:48:53] Maryrose: And we looked at mapping out how, mapping out how they used to do things, where all the kind of time consuming blockers are, and then how we could get the team to kind of use, you know, shared GPTs and, and automations and the like. So what they reported was coming into Dublin Tech Summit this year.

[00:49:11] Maryrose: Normally, it's a bit of a, uh, it's a struggle, you know, to run around and have everything done. They said that they had a 2 percent increase on their open rate of emails. Now their open rates were really high already. So two was pretty good. pretty good increase. But more importantly, they have, what were they saying?

[00:49:30] Maryrose: They have something like 46 social media channels for all of the events that they run. And what normally used to happen was that when one event would be on, like say Dublin Tech Summit, all the others would kind of suffer a little bit, the neglected children. So they basically went into Dublin Tech Summit this year with all 46 social media channels scheduled with evergreen content out the end of the year.

[00:49:52] Maryrose: I think that's amazing. And they're a smallish company, but they're just able to punch above their weight now because they're, they're working really [00:50:00] cohesively in, in the way that they're using their, the, the AI and that we brought in.

[00:50:07] Claire: I think, um, building on that, I mean, the time efficiency gain is the obvious. You know, ROI points that I think a lot of people can measure and feel when they're working with automations. Um, the automation that we've got on the hyper personalized emails actually has a layer to it where when an email sends from HubSpot, we have a, an automation that fires into Slack that tells us how that email has performed against.

[00:50:34] Claire: Specified benchmarks, and it just gives us a red, amber, green indicator, and so we're keeping a pulse the whole time on whenever those emails are firing. When do we see one that's an outlier, like it's performed exceptionally well, or it's dropped beneath the threshold. And just from that pulse check, um, you know, I know, anecdotally, that The hyper personalized automations we're sending are outperforming the non personalized one.

[00:50:58] Claire: That's how I know. So, we're trying to find ways to incorporate that ROI feedback loop, um, into the automation. So that, it's a, it's a full system. So it sends the email, but it's also reporting back on ROI for us so that we know it's worth doing. Um, yeah, it's a work in progress, but it's quite a fun project to close that loop.

[00:51:18] Amie: Yeah, it's interesting. I also feel like, um, people can get kind of like, depending on your mood, you get lazy with your email writing. So you might send a really personalized email out one day, and then maybe not so personalized the next. So it's kind of nice with, you know, the setup that you kind of have that.

[00:51:34] Amie: Repetitive, uh, um, care in the process. Um, since we only have a few minutes left, um, Mary Rhodes, I'd love to hear you talk a little bit more about what you called like waves of AI with marketing. Um, and you kind of been in AI way before many of us have even heard of it on Twitter. Uh, so you, you kind of have seen how, um, AI has changed the landscape of marketing.

[00:51:59] Amie: I would [00:52:00] love if you would share kind of your thoughts. Absolutely, I'd love to.

[00:52:04] Maryrose: Yeah, I was, uh, I've, I've always been very much an early adopter, future focused marketer. So I was kind of starting to use some of the early AI products like long before chat GPT 3. 5. The problem I had then was that, um, I, I would be telling clients and they didn't, they didn't want to have AI anywhere near their marketing work.

[00:52:24] Maryrose: Um, and then part of November, 2020. to happen and kind of changed everything. So the way I sort of see it is that there's kind of levels or stages or waves is really what the word I was using so I think wave zero was anything before 3. 5 landed like you might have seen a few things like say jasper or Um things like that and you might have been dabbling So wave, wave one is that first wave, which we've all just been through, which is where, you know, everybody's getting on to using chat to PT and Claude.

[00:52:56] Maryrose: And it's this exciting time where there's new features dropping. Uh, there's more powerful tools coming. We're, we're getting our heads around prompt engineering, um, all those kinds of big words that have been invented. I think we're kind of coming to the end. We are actually, I think at the end of that now where, you know, most, most marketers, anyway, they should have.

[00:53:16] Maryrose: You know, at least five to 10 GPTs that they have built, which is their best work, their best prompts, their most go to stuff that they use all the time. And now we're moving into this second wave, which is where we're taking, uh, entire pieces of our workflow and we're putting automations around it. And the thing that I really learned from working with Claire, when we're really brainstorming and what we put into this course was, I always think of like.

[00:53:42] Maryrose: You know, do the whole thing. And then I see myself, you know, living on a beat. But what Claire told me is that there's a lot more value in just taking, you know, small pieces that are like really, really, um, crying out for automation. And then. Doing lots and lots of [00:54:00] them and then ultimately connecting them.

[00:54:02] Maryrose: So that's kind of where we are at the moment. I'm urging everybody, everybody who's in the room, you know, if you haven't gone deep on, uh, bringing automation into your workflow, you really need to, because if you're a part of Ben's Bytes community, you're already beyond level one, wave one, you should be in wave two.

[00:54:20] Maryrose: And then the next bit will be wave three. And, you know, I think it's going to come much sooner than wave two. Wave 2 did. We don't know what it's going to contain, but I think if you're well set up with lots of automations, you'll be ready to kind of leap into whatever that new space is.

[00:54:37] Amie: Um, I might've opened Pandora's box with asking that question because now, uh, one last question in the chat is when you say every marketer should have at least five to 10 GPTs, um, can you give us just a high level overview of what those should look like and what those use cases should be?

[00:54:54] Maryrose: Sure. Well, I can just speak to the ones that we use ourselves.

[00:54:57] Maryrose: So we have, um, voice of us, which is, uh, as I said before, it's our own tone of voice. Then we have like your voice of the customer. So for example, you know, if you're a marketing consultant, you might have say two or three different clients. You know, you might have a pharma, you might have a wellness, you know, they're all going to have different tones of voice.

[00:55:16] Maryrose: So you want to have different GPTs set up for each one. Um, we also have, um, we call it salesy sales, um, which is, um, a beautiful one around helping you to, uh, use AI to identify new prospects. Um, we have an AI for sales course. I'm really lucky. See, I get to do all the courses. So we've been AI for sales course led by really cool creator called Danny O'Neill.

[00:55:41] Maryrose: And, um, I just, my favorite bit was that. So I created into a GPT that the team could use. Um, we also have a competitor analysis one, um, which is kind of derived from the thinkings of, um, do you know Liza Adams, I follow her a lot on LinkedIn. She does a lot of very kind of high [00:56:00] level strategic marketing stuff.

[00:56:01] Maryrose: So I kind of took one of the concepts that she had, and then I kind of built it out into a prompt that worked for us. And then once I had it working, I then built that into a GPT. So they're the ones that we have. When we're doing beginners courses for a wide range of sectors, like I always show them how to do things like RFPs, you know, upload every single successful tender response that you've ever done into a GPT.

[00:56:27] Maryrose: And then the next time you are doing a RFP, you can literally just drop a question in and it'll feed off your previously successful answers first. That, RFPs, tenders, um, if you're entering events or sorry, if you're entering awards, any of your previous award submissions, build them all into a GPT and then feed off that.

[00:56:48] Maryrose: That's a few of them.

[00:56:52] Ben: Yeah, it seems like it could go on and be endless and you can just keep making them over and over again. Um, but yeah, I just want to thank you and. This has been a great session and yeah, we will definitely upload everything or links and everything else to I'll get the recording done, edited and uploaded hopefully this evening and then everything will be connected there.

[00:57:10] Ben: But just want to thank you both so much for your time and glad that the demo gods did shine on us today and weren't too rough with us. Yeah, thank you so much, both. And thank you everyone for joining. And we'll see you in the next one.

[00:57:26] Maryrose: Thank you so much. See you again. Thanks so much. Bye. Bye.

‍