1 00:00:00,000 --> 00:00:09,960 Welcome to Artificially Intelligent Marketing, a weekly podcast where we stay on top of the 2 00:00:09,960 --> 00:00:15,700 latest trends, tips and tools in the world of marketing AI, helping you get the best 3 00:00:15,700 --> 00:00:18,520 results from your marketing efforts. 4 00:00:18,520 --> 00:00:23,040 Now let's join our hosts, Paul Avery and Martin Broadhurst. 5 00:00:23,040 --> 00:00:26,840 Welcome to episode 25 of Artificially Intelligent Marketing. 6 00:00:26,840 --> 00:00:32,480 Paul Avery here and this time I'm not joined by my colleague Martin Broadhurst because 7 00:00:32,480 --> 00:00:40,200 he is on tour at Inbound 2023 HubSpots marketing conference. 8 00:00:40,200 --> 00:00:44,920 And we'll hear from Martin at the end of the show as he tells us what he feels were his 9 00:00:44,920 --> 00:00:47,880 top takeaways, lots of generative AI in there. 10 00:00:47,880 --> 00:00:54,760 And we've also got a few interviews that Martin was able to record while on site at the show. 11 00:00:54,760 --> 00:01:01,080 So with us picking up from Martin later on, it's my job today to take you briefly through 12 00:01:01,080 --> 00:01:03,560 the AI news that we saw this week. 13 00:01:03,560 --> 00:01:06,280 And we're going to start with Slack. 14 00:01:06,280 --> 00:01:12,560 So Slack is introducing a new AI tool designed to help users navigate the workspace of Slack 15 00:01:12,560 --> 00:01:13,560 more efficiently. 16 00:01:13,560 --> 00:01:18,120 The tool is going to be called Slack AI and it's going to be tested this winter and it's 17 00:01:18,120 --> 00:01:23,160 going to offer features like summarizing threads and recapping channel highlights. 18 00:01:23,160 --> 00:01:28,520 It also brings a more advanced search function that effectively allows users to ask questions 19 00:01:28,520 --> 00:01:32,640 in natural language questions and then the tool is going to provide a summary against 20 00:01:32,640 --> 00:01:38,100 those questions and then links to any relevant conversations from Slack. 21 00:01:38,100 --> 00:01:43,720 In addition to launching Slack AI, Slack is also going to launch Slack Lists, which is 22 00:01:43,720 --> 00:01:47,600 a feature to help users track projects and tasks within their team. 23 00:01:47,600 --> 00:01:52,920 A little bit like what you'll see if you're using project management apps like Asana or 24 00:01:52,920 --> 00:01:56,880 if you're used to using database apps like Airtable. 25 00:01:56,880 --> 00:02:02,960 Slack is also enhancing its integration capabilities with a new workflow builder where the aim 26 00:02:02,960 --> 00:02:09,200 is to simplify the addition of third party apps into your Slack ecosystem. 27 00:02:09,200 --> 00:02:16,240 So this news comes as a follow up to previous efforts by Slack to leverage AI with the introduction 28 00:02:16,240 --> 00:02:21,440 of their chat QPT bot for assisting in things like message replies. 29 00:02:21,440 --> 00:02:28,440 And so ultimately this is the ongoing evolution of Slack as a tool baking AI to try and make 30 00:02:28,440 --> 00:02:30,720 things easier for its users. 31 00:02:30,720 --> 00:02:35,680 If you're a marketer in a marketing team that uses Slack, then hopefully these new features 32 00:02:35,680 --> 00:02:39,120 are going to make it easier for your team to collaborate. 33 00:02:39,120 --> 00:02:44,080 Certainly the ability to access summarized conversations is going to aid in more efficient 34 00:02:44,080 --> 00:02:49,680 communication and those Slack Lists may help you better with project management. 35 00:02:49,680 --> 00:02:53,360 So I think this is one of those things that if you're a Slack user, jump on in, have a 36 00:02:53,360 --> 00:02:59,020 little play once we start to see this emerge at the back end of this year and see if it 37 00:02:59,020 --> 00:03:01,080 improves your experience. 38 00:03:01,080 --> 00:03:05,720 Is it worth adopting Slack because of this new tool? 39 00:03:05,720 --> 00:03:07,760 My suggestion would probably be not. 40 00:03:07,760 --> 00:03:12,240 But of course Slack has a number of other benefits and if you were considering the tool, 41 00:03:12,240 --> 00:03:16,960 then this might be the one that nudged you over and made you think, oh yeah, we'll give 42 00:03:16,960 --> 00:03:18,360 that one a go. 43 00:03:18,360 --> 00:03:22,840 So moving on from Slack, now we're going to talk a bit about Anthropic. 44 00:03:22,840 --> 00:03:27,520 So we've talked a fair bit about Anthropic on the podcast in the past. 45 00:03:27,520 --> 00:03:33,840 They are the developers of the Afflord Chatbot, which is a competitor to ChatGPT. 46 00:03:33,840 --> 00:03:39,560 And in essence, what Anthropic have done is roll out a pay plan for Claw 2, which is going 47 00:03:39,560 --> 00:03:43,000 to be available in the US and the UK to start with. 48 00:03:43,000 --> 00:03:46,480 Now though, people who've listened to the podcast know that there's a few things that 49 00:03:46,480 --> 00:03:49,440 we love about Claw 2 compared to ChatGPT. 50 00:03:49,440 --> 00:03:54,240 So for example, Claw 2 has a much larger context window of up to around 75,000 words. 51 00:03:54,240 --> 00:03:59,440 So there's lots of cool stuff that you can do with it that's much easier than with ChatGPT, 52 00:03:59,440 --> 00:04:02,000 like having it summarize large documents. 53 00:04:02,000 --> 00:04:08,160 It also produces its outputs nice and quickly and Claw 2, since it was released, has shown 54 00:04:08,160 --> 00:04:10,560 that it's much better at more complex reasoning. 55 00:04:10,560 --> 00:04:15,520 A couple of episodes ago, Martin went so far as to say that it's actually now part of his 56 00:04:15,520 --> 00:04:17,600 go-to workflow. 57 00:04:17,600 --> 00:04:22,640 So as a power user, he's probably going to need this new Clawed Pro plan because it's 58 00:04:22,640 --> 00:04:30,400 going to offer five times more usage than just the free part at a monthly fee of $20. 59 00:04:30,400 --> 00:04:36,680 So in essence, if you pay for Clawed Pro, you're going to be able to use it a lot more 60 00:04:36,680 --> 00:04:40,520 without maxing out your capacity within the tool. 61 00:04:40,520 --> 00:04:46,040 You're also going to get priority during peak times and Anthropic promises early access 62 00:04:46,040 --> 00:04:47,560 to upcoming features. 63 00:04:47,560 --> 00:04:52,920 So it's quite interesting to see this from Anthropic with them positioning this premium 64 00:04:52,920 --> 00:04:59,200 offering of Claw 2 is basically them competing pretty much directly with OpenAI's ChatGPT+, 65 00:04:59,200 --> 00:05:01,680 which is the paid version. 66 00:05:01,680 --> 00:05:03,280 So this is quite interesting to see. 67 00:05:03,280 --> 00:05:08,000 I think as marketers, the introduction of Claw Pro is part of this wider battle of the 68 00:05:08,000 --> 00:05:09,000 bots, right? 69 00:05:09,000 --> 00:05:15,360 So the key question remains, will Claw 2 carve out its niche in a market where ChatGPT+, 70 00:05:15,360 --> 00:05:18,160 already has a major stronghold? 71 00:05:18,160 --> 00:05:22,160 We've also talked in the past about the release of Google Duet. 72 00:05:22,160 --> 00:05:28,760 So having generative AI tools baked into Gmail, Google Slides, Google Sheets, et cetera. 73 00:05:28,760 --> 00:05:35,240 And there are a number of companies currently testing Windows Copilot and Office 365 Copilot. 74 00:05:35,240 --> 00:05:39,160 And that should be released fairly soon to the rest of us as well. 75 00:05:39,160 --> 00:05:45,160 And of course, that brings generative AI into the tools that people use every day like PowerPoint, 76 00:05:45,160 --> 00:05:49,040 Word, Gmail, Sheets, and all of that good stuff. 77 00:05:49,040 --> 00:05:55,840 So this is just another story in the rapidly evolving space of generative AI tools for 78 00:05:55,840 --> 00:05:58,960 brainstorming, reasoning, content creation, et cetera. 79 00:05:58,960 --> 00:06:03,840 And I think as a marketer, it's just about staying on top of all these different news 80 00:06:03,840 --> 00:06:08,140 items so that you can get a feel for what the landscape looks like, what tools might 81 00:06:08,140 --> 00:06:14,780 be best for different applications and use cases, and just the types of pricing as well. 82 00:06:14,780 --> 00:06:21,360 Because I think seeing the price of Claw 2 be similar to ChatGPT is not a surprise. 83 00:06:21,360 --> 00:06:26,440 And in many ways, maybe ChatGPT+, has set the bar there for what people are happy to 84 00:06:26,440 --> 00:06:29,720 play or pay in order to access these tools. 85 00:06:29,720 --> 00:06:33,720 And we may see others follow suit there. 86 00:06:33,720 --> 00:06:36,680 The next story is about Microsoft. 87 00:06:36,680 --> 00:06:41,440 And according to the AI newsletter, the neuromaniac of our favorite newsletters, Microsoft has 88 00:06:41,440 --> 00:06:47,120 stated that they'll shoulder any legal expenses if your business encounters legal challenges 89 00:06:47,120 --> 00:06:50,000 from using their AI offerings. 90 00:06:50,000 --> 00:06:53,560 If this sounds a bit familiar to you, then that's probably not a surprise because Adobe 91 00:06:53,560 --> 00:06:59,900 has made a comparable commitment in the past for its AI tool Firefly. 92 00:06:59,900 --> 00:07:05,440 So the backdrop here is that a number of AI companies, including Mid Journey, GitHub Copilot, 93 00:07:05,440 --> 00:07:10,920 OpenAI, have all faced legal actions for allegedly utilizing creative content to train their 94 00:07:10,920 --> 00:07:14,200 models without proper permissions. 95 00:07:14,200 --> 00:07:19,720 And as you might expect, the scenarios made some businesses hesitant to use the tools 96 00:07:19,720 --> 00:07:25,200 because of any potential issues that they then may find themselves in if they're using 97 00:07:25,200 --> 00:07:26,720 them. 98 00:07:26,720 --> 00:07:33,520 But when you see Microsoft make a move like this, it's another interesting attempt to 99 00:07:33,520 --> 00:07:39,000 really give customers confidence that they can trust the tools that they're using, not 100 00:07:39,000 --> 00:07:45,080 just in terms of how the tools were trained, but also in terms of what data that you give 101 00:07:45,080 --> 00:07:49,320 them and how much you can trust the information that you're being put in won't be used to 102 00:07:49,320 --> 00:07:54,700 further train the models or for other purposes that you haven't signed up to. 103 00:07:54,700 --> 00:08:01,020 And ultimately, these types of moves are mostly, I think, targeted at large enterprise customers 104 00:08:01,020 --> 00:08:09,220 who might not have had confidence to step in and use tools like ChatGPT and other tools. 105 00:08:09,220 --> 00:08:15,320 I think with the release of ChatGPT Enterprise and now Microsoft promising to cover legal 106 00:08:15,320 --> 00:08:21,880 costs if you get into hot water, the generative AI battle is on to attract high value enterprise 107 00:08:21,880 --> 00:08:22,880 customers. 108 00:08:22,880 --> 00:08:27,920 If you work at one of those sort of larger companies or maybe more risk averse companies 109 00:08:27,920 --> 00:08:32,160 and you've been itching to get your hands on these types of generative AI tools at work, 110 00:08:32,160 --> 00:08:36,920 then your wait might soon be over. 111 00:08:36,920 --> 00:08:44,200 Next we're talking Zoom and the popular digital video meeting and webinar software. 112 00:08:44,200 --> 00:08:47,300 And Zoom has rolled out its AI companion. 113 00:08:47,300 --> 00:08:52,000 So the new assistant previously known as Zoom IQ offers a range of functionalities that 114 00:08:52,000 --> 00:08:57,360 include things like if you're late to a meeting, it will provide you a quick written recap. 115 00:08:57,360 --> 00:09:01,920 It's very good at summarizing meetings, pinpointing key discussions and outlining actionable next 116 00:09:01,920 --> 00:09:02,920 steps. 117 00:09:02,920 --> 00:09:09,160 And it can even assist users in crafting chat responses in the Zoom chat tool. 118 00:09:09,160 --> 00:09:13,860 And there's more that Zoom promises will be coming to the tool because it promises that 119 00:09:13,860 --> 00:09:19,280 at some point it will aid in drafting emails, offer chat message summaries, recognize meeting 120 00:09:19,280 --> 00:09:25,240 intents in chat conversations and even support brainstorming on a digital whiteboard. 121 00:09:25,240 --> 00:09:30,120 Coolest part about all of this is that Zoom is going to opt for this to paid subscribers 122 00:09:30,120 --> 00:09:33,280 with no additional charge. 123 00:09:33,280 --> 00:09:34,280 This is fairly awesome. 124 00:09:34,280 --> 00:09:35,280 I'm a user of Loom. 125 00:09:35,280 --> 00:09:44,520 I love Loom for screen recordings and they've been beta testing AI driven title creation, 126 00:09:44,520 --> 00:09:47,160 chapters for videos, video summaries. 127 00:09:47,160 --> 00:09:48,160 And I found it pretty useful. 128 00:09:48,160 --> 00:09:49,160 It's a bit hit and miss. 129 00:09:49,160 --> 00:09:50,160 I think it gets things wrong. 130 00:09:50,160 --> 00:09:55,840 And I thought this was a really nice addition to the tool that I really, it's kind of cool. 131 00:09:55,840 --> 00:10:00,240 I think it's a bit buggy so that I'm not entirely sure I'd trust it commercially. 132 00:10:00,240 --> 00:10:06,760 But then Loom said this week or last week that you now need to pay if you want to access 133 00:10:06,760 --> 00:10:07,760 Loom AI. 134 00:10:07,760 --> 00:10:08,760 And I understand, right? 135 00:10:08,760 --> 00:10:15,800 It costs money to access AI driven tools and these companies have to make that money back 136 00:10:15,800 --> 00:10:19,920 somewhere but I think it's very, very interesting that Loom are looking to charge users. 137 00:10:19,920 --> 00:10:22,160 I certainly won't be paying for the AI tool. 138 00:10:22,160 --> 00:10:24,840 Honestly, I don't think it's worth the extra money. 139 00:10:24,840 --> 00:10:28,200 But Zoom is saying you will have it for free. 140 00:10:28,200 --> 00:10:34,680 We talked a little bit earlier about ChatGPT and Claude Pro sort of setting a market expectation 141 00:10:34,680 --> 00:10:40,060 of paying $20 a month for that type of generative AI experience. 142 00:10:40,060 --> 00:10:45,680 If Zoom are going to offer AI driven tools as part of their existing subscriptions without 143 00:10:45,680 --> 00:10:49,760 increasing their subscription fee, which of course there's no guarantees that won't come 144 00:10:49,760 --> 00:10:50,760 later down the line. 145 00:10:50,760 --> 00:10:55,040 But if that was the case, it would be a very interesting pricing bar to set that if you're 146 00:10:55,040 --> 00:11:04,100 a meeting tool like Teams, Google Me, then for the meeting driven benefits on their own, 147 00:11:04,100 --> 00:11:08,880 you probably can't afford to charge much extra because people can get that cool AI wizardry 148 00:11:08,880 --> 00:11:13,920 within Zoom as part of their existing subscriptions. 149 00:11:13,920 --> 00:11:18,360 If you're a marketer and you use Zoom either for webinars or for your internal meetings, 150 00:11:18,360 --> 00:11:24,040 then of course all of the features and benefits that we've discussed in this particular news 151 00:11:24,040 --> 00:11:28,160 story so far will be a benefit to you. 152 00:11:28,160 --> 00:11:33,560 And if you're already using additional paid tools like Otter or Tacti to record meeting 153 00:11:33,560 --> 00:11:38,920 transcriptions and generate AI driven summaries, it might be possible that you can just go 154 00:11:38,920 --> 00:11:42,600 all in on Zoom and drop some of those other tools you've been using. 155 00:11:42,600 --> 00:11:47,060 So again, another very interesting space to watch. 156 00:11:47,060 --> 00:11:50,400 The next story is about Sprigg. 157 00:11:50,400 --> 00:11:55,320 Now Sprigg is known for its in-app surveys and Sprigg has launched a new feature named 158 00:11:55,320 --> 00:11:57,640 AI Analysis for Surveys. 159 00:11:57,640 --> 00:12:04,840 So this tool uses generative AI to delve into survey data and offer immediate insights. 160 00:12:04,840 --> 00:12:09,800 What sets it apart is its ability to turn survey data into a conversational AI interface. 161 00:12:09,800 --> 00:12:14,760 So in effect what you can do is you can directly ask your survey results questions in a natural 162 00:12:14,760 --> 00:12:20,520 language chatbot style way and the AI will sift through the data including qualitative 163 00:12:20,520 --> 00:12:27,280 responses so long form text responses potentially to provide relevant insights. 164 00:12:27,280 --> 00:12:32,280 Ryan Glasgow, which is Sprigg's CEO, highlighted that users can pose custom questions to their 165 00:12:32,280 --> 00:12:38,720 survey data and the AI will analyze the entirety of the responses to provide answers and in 166 00:12:38,720 --> 00:12:43,440 a move to make their tools more accessible, Sprigg has broadened its free plan now encompassing 167 00:12:43,440 --> 00:12:48,160 in-product surveys, session replays and open text AI analysis. 168 00:12:48,160 --> 00:12:55,080 So if you do lots of surveys, either for bus to repair market research, messaging creation, 169 00:12:55,080 --> 00:12:58,600 customer analysis, I think it's probably worth going and have a little look at what Sprigg 170 00:12:58,600 --> 00:13:03,980 are doing because you might find that this new AI driven analysis tool is beneficial 171 00:13:03,980 --> 00:13:05,200 for you. 172 00:13:05,200 --> 00:13:10,080 Especially given that traditional survey analysis can be quite labor intensive, especially if 173 00:13:10,080 --> 00:13:15,120 you are using surveys that have those open-ended text answers, it would be interesting to see 174 00:13:15,120 --> 00:13:21,760 if when you're playing with Sprigg's AI analysis for surveys, if you can really pull out insights 175 00:13:21,760 --> 00:13:26,300 and trends for that information that you then double check manually and just see how good 176 00:13:26,300 --> 00:13:31,660 the AI is because if it's good, could really change how market research is done, how customer 177 00:13:31,660 --> 00:13:33,660 sentiment analysis is done. 178 00:13:33,660 --> 00:13:38,240 And of course, if we're seeing this type of thing being released by Sprigg, then expect 179 00:13:38,240 --> 00:13:44,760 to see it perhaps in other tools like Typeform or SurveyMonkey or other other tools that 180 00:13:44,760 --> 00:13:46,680 you can use for market research. 181 00:13:46,680 --> 00:13:49,800 I know I've not used Sprigg before, but I think I'm going to go and explore that and 182 00:13:49,800 --> 00:13:53,000 see how powerful it is. 183 00:13:53,000 --> 00:13:58,400 So the next story is not strictly a marketing story, but it was just too cool to not include 184 00:13:58,400 --> 00:13:59,400 and it's from Amazon. 185 00:13:59,400 --> 00:14:03,640 So this week, Amazon introduced a new technology called Amazon One, allowing customers to 186 00:14:03,640 --> 00:14:10,440 use their palm for activities like payments, loyalty card presentation, and page verification. 187 00:14:10,440 --> 00:14:16,320 Create the system, Amazon utilized generative AI to produce millions of synthetic palm images, 188 00:14:16,320 --> 00:14:18,200 a method known as synthetic data. 189 00:14:18,200 --> 00:14:24,200 And this approach enabled Amazon to train their model effectively, achieving near 100% 190 00:14:24,200 --> 00:14:25,200 accuracy. 191 00:14:25,200 --> 00:14:30,120 So in essence, you can imagine going into a shop and you don't need your card, you don't 192 00:14:30,120 --> 00:14:34,920 need your phone, you don't need your watch, you just hold your palm above a scanner and 193 00:14:34,920 --> 00:14:36,400 that's how you pay. 194 00:14:36,400 --> 00:14:42,560 So you go out pretty much with nothing and you can still pay just using your palm. 195 00:14:42,560 --> 00:14:47,200 Amazon has interestingly emphasized the system's privacy features. 196 00:14:47,200 --> 00:14:51,560 So Amazon One operates beyond the usual light spectrum, so it can't recognize things like 197 00:14:51,560 --> 00:14:53,440 gender or skin tone. 198 00:14:53,440 --> 00:14:59,040 And the primary purpose, Amazon says, is to match a unique identity with a payment method, 199 00:14:59,040 --> 00:15:05,720 not to be able to identify individuals in any way outside of that, you know, asterisk, 200 00:15:05,720 --> 00:15:06,720 asterisk. 201 00:15:06,720 --> 00:15:11,560 So I can see that there might be possible privacy issues here, whether this technology 202 00:15:11,560 --> 00:15:14,680 actually takes off or not is going to be an interesting one. 203 00:15:14,680 --> 00:15:19,440 There's lots of companies working on the power of things like retinal scans to provide this 204 00:15:19,440 --> 00:15:21,920 same sort of capability. 205 00:15:21,920 --> 00:15:28,320 But obviously having your retina scanned is potentially a bit more technically challenging 206 00:15:28,320 --> 00:15:33,160 and also not a great user experience compared to just waving your palm over something. 207 00:15:33,160 --> 00:15:37,520 So yeah, not technically a marketing story, but wanted to include this week because it 208 00:15:37,520 --> 00:15:40,320 sounded really cool. 209 00:15:40,320 --> 00:15:46,720 Next story is that you can now access a new Canva plugin if you are a ChatGPT Plus user, 210 00:15:46,720 --> 00:15:52,360 which makes it simpler to create visuals such as logos and social media banners. 211 00:15:52,360 --> 00:15:57,000 Now before this integration, to combine the powers of Canva and ChatGPT was a bit of a 212 00:15:57,000 --> 00:16:00,200 task and with the plugin, it's a bit more straightforward. 213 00:16:00,200 --> 00:16:04,920 How it works is you go into ChatGPT, you go to the plugin store, you install the Canva 214 00:16:04,920 --> 00:16:08,840 plugin and then you select it when you are adding your prompt. 215 00:16:08,840 --> 00:16:13,120 You describe the visual you want in the chat box, prompt box, and then ChatGPT will present 216 00:16:13,120 --> 00:16:18,160 some design options to you, which in my experience playing with the tool has been based on Canva's 217 00:16:18,160 --> 00:16:20,260 existing templates. 218 00:16:20,260 --> 00:16:24,480 You choose one and then you'll be redirected to Canva to go make the edits that you need 219 00:16:24,480 --> 00:16:28,160 to make and then you can download your Fav Canva. 220 00:16:28,160 --> 00:16:32,800 As this is a plugin, you can obviously only access it if you're a ChatGPT Plus subscriber 221 00:16:32,800 --> 00:16:38,320 at $20 a month because you don't get plugin access with the free version of ChatGPT. 222 00:16:38,320 --> 00:16:39,680 What does this mean for marketers? 223 00:16:39,680 --> 00:16:45,360 Well, it could potentially streamline your ChatGPT Canva design process. 224 00:16:45,360 --> 00:16:49,840 So you could probably end up with slightly faster production workflows, although I have 225 00:16:49,840 --> 00:16:55,240 to say in my quick tests, it mostly acted like a smart search for finding good templates 226 00:16:55,240 --> 00:16:57,080 among all the options available. 227 00:16:57,080 --> 00:17:00,720 And let's be honest, Canva does have huge amounts of templates and it's sometimes a 228 00:17:00,720 --> 00:17:04,520 bit for paying searching or scrolling through them looking for a good one. 229 00:17:04,520 --> 00:17:08,360 But in essence, you use ChatGPT, you say, look, I needed templates going to be perfect 230 00:17:08,360 --> 00:17:09,640 for this use case. 231 00:17:09,640 --> 00:17:11,960 And then it surfaces three for you to choose from. 232 00:17:11,960 --> 00:17:15,520 And if you don't like them, then you can ask it to show you some more and you can give 233 00:17:15,520 --> 00:17:19,680 some direction on, oh no, these were a bit too much like A and what I was really looking 234 00:17:19,680 --> 00:17:20,680 for was B. 235 00:17:20,680 --> 00:17:26,320 So I think it may be, at least in my early tests, looks like a fairly powerful quick 236 00:17:26,320 --> 00:17:29,000 search for Canva templates. 237 00:17:29,000 --> 00:17:34,080 But certainly I couldn't get it to do anything really interesting, like starting to influence 238 00:17:34,080 --> 00:17:36,680 the design from within ChatGPT. 239 00:17:36,680 --> 00:17:41,020 I had to go edit the template directly in Canva. 240 00:17:41,020 --> 00:17:46,320 So kind of cool, maybe not quite as exciting as it sounds on the surface. 241 00:17:46,320 --> 00:17:47,600 Right. 242 00:17:47,600 --> 00:17:51,640 The last story this week is about Gizmodo. 243 00:17:51,640 --> 00:17:56,520 So Gizmodo is a tech publication that's owned by Geo Media. 244 00:17:56,520 --> 00:18:01,760 And this week, Geo Media has decided to lay off the editors in its Spanish language site, 245 00:18:01,760 --> 00:18:04,160 Gizmodo en Espanol. 246 00:18:04,160 --> 00:18:08,760 In place of the human editors, they're now leaning on AI to handle translations. 247 00:18:08,760 --> 00:18:14,640 So before this change, Gizmodo en Espanol had a dedicated team producing original content 248 00:18:14,640 --> 00:18:18,480 and also adapting English articles for their Spanish readers. 249 00:18:18,480 --> 00:18:23,480 But now articles on the site carry a disclaimer indicating that they've been translated by 250 00:18:23,480 --> 00:18:26,720 AI, which might need to dislike content variations. 251 00:18:26,720 --> 00:18:31,800 So in essence, no custom content is being created for the Spanish readers now. 252 00:18:31,800 --> 00:18:36,560 It's all being translated directly from any English content that's being created by those 253 00:18:36,560 --> 00:18:39,160 teams within Gizmodo. 254 00:18:39,160 --> 00:18:44,960 As you can imagine, in the early days of why we figure out these tools, it hasn't gone 255 00:18:44,960 --> 00:18:45,960 flawlessly. 256 00:18:45,960 --> 00:18:46,960 There's been some hiccups. 257 00:18:46,960 --> 00:18:50,880 So some readers have noticed that articles can start in Spanish and then unexpectedly 258 00:18:50,880 --> 00:18:53,320 switch to English. 259 00:18:53,320 --> 00:18:57,200 And it's also worth noting that Geo Media's initial foray into AI written articles in 260 00:18:57,200 --> 00:19:00,920 July encountered issues with factual inaccuracies. 261 00:19:00,920 --> 00:19:06,800 So a lot of the issues that we know with generative AI very much front and center here. 262 00:19:06,800 --> 00:19:10,840 The GMG Union, which represents the writers, as you would imagine, avoids their concerns 263 00:19:10,840 --> 00:19:12,180 over this move. 264 00:19:12,180 --> 00:19:17,320 They highter that Geo Media has previously suggested that AI would complement human writers, 265 00:19:17,320 --> 00:19:18,640 not replace them. 266 00:19:18,640 --> 00:19:23,400 The Union also pointed out the uncertain future faced by the site's employees. 267 00:19:23,400 --> 00:19:25,400 So what does this mean for us as marketers? 268 00:19:25,400 --> 00:19:28,240 Well, of course, there's multiple levels here. 269 00:19:28,240 --> 00:19:36,160 Gizmodo is not the first publication to explore or use generative AI to produce content. 270 00:19:36,160 --> 00:19:39,400 Most of the examples we get to hear about in the news are usually when it goes wrong 271 00:19:39,400 --> 00:19:43,800 for obvious reasons, but certainly we do know that other publications have been doing some 272 00:19:43,800 --> 00:19:47,360 of this and I suspect many more are. 273 00:19:47,360 --> 00:19:52,760 But I think what's really interesting for us as marketers is that this is another sign 274 00:19:52,760 --> 00:19:55,520 of that changing media landscape. 275 00:19:55,520 --> 00:20:00,280 So if you're a PR professional, for example, we're really going to have to think about 276 00:20:00,280 --> 00:20:04,480 how we adapt to a world where your press releases and pitches might be read and translated by 277 00:20:04,480 --> 00:20:10,120 AI before they even reach the human journalist, if indeed they ever do. 278 00:20:10,120 --> 00:20:13,840 And then the precision of the language that they're using in our press releases, etc., 279 00:20:13,840 --> 00:20:15,000 becomes paramount, right? 280 00:20:15,000 --> 00:20:20,840 Because any ambiguity or cultural nuance might be lost or misinterpreted by an AI tool, which 281 00:20:20,840 --> 00:20:23,600 could lead to miscommunication or missed opportunities. 282 00:20:23,600 --> 00:20:27,840 So I think that's just a risk for the publications, but it's also a risk for us. 283 00:20:27,840 --> 00:20:31,680 Secondly, building relationships is at the heart of PR. 284 00:20:31,680 --> 00:20:37,080 If we're seeing fewer human editors that are themselves being replaced, sadly, in some 285 00:20:37,080 --> 00:20:41,800 cases by AI, then the traditional approach of cultivating relationships with journalists 286 00:20:41,800 --> 00:20:47,800 and editors is both much more important because you really need to identify those high-profile 287 00:20:47,800 --> 00:20:53,600 journalists and editors in your industry that have such a powerful voice that it would be 288 00:20:53,600 --> 00:20:59,120 folly for their publications to try and replace them with AI and build strong relationships 289 00:20:59,120 --> 00:21:01,120 with those people. 290 00:21:01,120 --> 00:21:07,080 But we also need to understand how the media are using technology probably more than ever 291 00:21:07,080 --> 00:21:08,080 before. 292 00:21:08,080 --> 00:21:13,240 And I think one thing's for sure, the more publications lean into AI-generated content, 293 00:21:13,240 --> 00:21:17,940 we might see that their appetite for things like contributed editorial from companies 294 00:21:17,940 --> 00:21:22,160 or even our press releases that we send them, it could reduce over time, right? 295 00:21:22,160 --> 00:21:26,800 Because they might not need to rely on those interactions with companies as much as they 296 00:21:26,800 --> 00:21:29,680 can just leverage generative AI to produce content. 297 00:21:29,680 --> 00:21:35,800 Of course, we can have a very long conversation about how good that AI-generated content is 298 00:21:35,800 --> 00:21:37,480 going to be. 299 00:21:37,480 --> 00:21:43,000 But if that ability for us to add value to the publishers through contributed editorial 300 00:21:43,000 --> 00:21:50,920 is lessened, we may see that earned media opportunities dry up even more and are replaced 301 00:21:50,920 --> 00:21:55,800 by pay-for-play or pay-for-appearance opportunities instead. 302 00:21:55,800 --> 00:22:01,120 So I think there's a layer underneath all of this, whereas a marketer or a communicator 303 00:22:01,120 --> 00:22:05,680 is a PR pro especially, we have to look at what's going on in the media and really think 304 00:22:05,680 --> 00:22:12,840 how is this going to really impact on our approach to PR and media relations. 305 00:22:12,840 --> 00:22:14,920 So that's all the stories for this week. 306 00:22:14,920 --> 00:22:16,560 Martin's going to be back in the UK. 307 00:22:16,560 --> 00:22:21,320 In fact, I think he's landing about now and he'll be with me again on the podcast next 308 00:22:21,320 --> 00:22:22,320 week. 309 00:22:22,320 --> 00:22:25,640 So with that being said, we're going to segue into learning a little bit more about what 310 00:22:25,640 --> 00:22:30,480 Martin found out in Bound 23 in Boston this week. 311 00:22:30,480 --> 00:22:32,480 Hello, everyone. 312 00:22:32,480 --> 00:22:39,840 This is Martin coming to you from Boston, where I've just attended the HubSpot conference 313 00:22:39,840 --> 00:22:42,580 in Bound 23. 314 00:22:42,580 --> 00:22:50,120 This was a conference packed full of AI right from the get-go, to be honest. 315 00:22:50,120 --> 00:22:56,880 Yamini Rangan, the CEO of HubSpot, opened with a keynote that really set the tone for 316 00:22:56,880 --> 00:22:58,400 what we could expect. 317 00:22:58,400 --> 00:23:03,760 Yamini honed in on the major tech changes, particularly generative AI, entering the picture 318 00:23:03,760 --> 00:23:06,080 in a big way over the past year. 319 00:23:06,080 --> 00:23:11,880 And she claims that we're stepping into an age of intelligence where AI isn't just data 320 00:23:11,880 --> 00:23:13,280 crunching anymore. 321 00:23:13,280 --> 00:23:20,280 Instead, AI is about generating new content, new insights, and importantly for an organization 322 00:23:20,280 --> 00:23:25,600 like HubSpot and its customers, new experiences. 323 00:23:25,600 --> 00:23:28,800 For businesses that get this right, the rewards are massive. 324 00:23:28,800 --> 00:23:33,880 She spoke about some research they'd done at HubSpot, which identified the fastest growing 325 00:23:33,880 --> 00:23:36,040 companies and what made them different. 326 00:23:36,040 --> 00:23:41,920 And it was companies that were generating the best customer connections. 327 00:23:41,920 --> 00:23:47,120 She said they're growing five times faster than their competitors. 328 00:23:47,120 --> 00:23:50,800 Her advice was to start small, but act fast. 329 00:23:50,800 --> 00:23:55,080 Some practical tips that she gave were to get chatbots added onto the website. 330 00:23:55,080 --> 00:24:00,840 If you can get AI-powered chatbots resolving customer queries, it can make a big impact, 331 00:24:00,840 --> 00:24:09,920 so much so that HubSpot's own AI-powered chatbot is now successfully resolving 78% questions 332 00:24:09,920 --> 00:24:12,160 it receives automatically. 333 00:24:12,160 --> 00:24:16,000 She also recommended that you pivot your content strategy. 334 00:24:16,000 --> 00:24:23,960 So she's talking about creating more content and more experiences with AI helping the distribution, 335 00:24:23,960 --> 00:24:26,160 but not necessarily just the generation. 336 00:24:26,160 --> 00:24:34,140 You need human expertise, podcasts, video content, audio content, with AI assisting 337 00:24:34,140 --> 00:24:38,680 the production, but not the sole driver of it. 338 00:24:38,680 --> 00:24:45,200 Let's just switch gears slightly and have a look at what Dharmesh Shah, who was one 339 00:24:45,200 --> 00:24:52,920 of the co-founders of HubSpot, I think he's the chief technical officer still. 340 00:24:52,920 --> 00:24:59,840 This is a guy that fell in love with generative AI very early on. 341 00:24:59,840 --> 00:25:07,540 In fact, he said that he even had some intriguing conversations with OpenAI CEO Sam Altman 342 00:25:07,540 --> 00:25:13,840 when GPT-3 launched and he managed to get early access to the model. 343 00:25:13,840 --> 00:25:18,480 Another interesting tidbit he said was that he's an early investor in OpenAI. 344 00:25:18,480 --> 00:25:24,600 I'm sure that's going to pay dividends for him personally. 345 00:25:24,600 --> 00:25:30,400 He highlighted the McKinsey report that said that generative AI is poised to add between 346 00:25:30,400 --> 00:25:39,480 2.6 trillion to 4 trillion annually across all sectors and that report highlighted sales, 347 00:25:39,480 --> 00:25:45,760 marketing, and customer service as being the business functions where generative AI will 348 00:25:45,760 --> 00:25:48,840 make the biggest impact. 349 00:25:48,840 --> 00:25:54,520 As HubSpot has sales, marketing, and customer service hubs in its core offering, you can 350 00:25:54,520 --> 00:25:57,840 see why they might be excited about this. 351 00:25:57,840 --> 00:26:05,880 So Darmesh urged everyone to get their hands on next-gen CRMs that are full of rich data 352 00:26:05,880 --> 00:26:09,000 but driven by AI. 353 00:26:09,000 --> 00:26:15,280 Getting these insights pulled out of the data so that you're able to deliver better customer 354 00:26:15,280 --> 00:26:16,280 experiences. 355 00:26:16,280 --> 00:26:22,400 They think of it as AI and data on steroids working on tandem with human skills, whether 356 00:26:22,400 --> 00:26:29,800 it's sales reps, marketers, or customer support reps delivering highly personalized experiences 357 00:26:29,800 --> 00:26:30,800 for the customers. 358 00:26:30,800 --> 00:26:36,160 Now let's get into some of the nitty gritty on the actual product updates that HubSpot 359 00:26:36,160 --> 00:26:37,160 is rolling out. 360 00:26:37,160 --> 00:26:42,960 If you are a HubSpot customer, there are some really exciting developments here. 361 00:26:42,960 --> 00:26:47,280 The first thing I want to say is that HubSpot AI is being baked into everything. 362 00:26:47,280 --> 00:26:50,960 It's not just a sprinkling of AI here and there. 363 00:26:50,960 --> 00:26:55,640 This is a full-on integration of AI across the HubSpot product line. 364 00:26:55,640 --> 00:27:01,820 I think we saw that from the roadmap that they had published on the website. 365 00:27:01,820 --> 00:27:09,320 Everything from understanding customer content, company data, it's promising to bring forward 366 00:27:09,320 --> 00:27:14,000 new levels of automation to the system. 367 00:27:14,000 --> 00:27:19,240 So back to what Yaroni said earlier, she was talking about bringing chat bots onto your 368 00:27:19,240 --> 00:27:20,240 website. 369 00:27:20,240 --> 00:27:21,920 These are the tools that they're bringing forward. 370 00:27:21,920 --> 00:27:29,720 They're also doing prediction models and we should see in the future, virtual agents working 371 00:27:29,720 --> 00:27:35,720 seamlessly to make your life as a marketer or a sales rep or whatever your role may be 372 00:27:35,720 --> 00:27:38,160 a little bit easier. 373 00:27:38,160 --> 00:27:44,440 So getting into the specifics of it, they've announced a content assistant image creation 374 00:27:44,440 --> 00:27:45,440 tool. 375 00:27:45,440 --> 00:27:47,800 So the content assistant already existed. 376 00:27:47,800 --> 00:27:50,040 You could sign up and get it in beta. 377 00:27:50,040 --> 00:27:52,920 But now they've added image creation to this. 378 00:27:52,920 --> 00:27:59,960 I think it's running on Dali, but I'm not entirely sure that they didn't or at least 379 00:27:59,960 --> 00:28:05,600 I didn't hear the specifics as to which model they were actually using. 380 00:28:05,600 --> 00:28:07,520 So I mentioned content assistant. 381 00:28:07,520 --> 00:28:13,440 They've also launched campaign assistant so you can input your goals, your audience and 382 00:28:13,440 --> 00:28:22,280 your brand voice and then let AI whip up the whole marketing campaign, which is kind of 383 00:28:22,280 --> 00:28:24,960 accurate but not wholly. 384 00:28:24,960 --> 00:28:30,800 Right now it doesn't automatically generate all of the assets, but in discussions that 385 00:28:30,800 --> 00:28:37,200 I had with the product team, it seems like the longer term vision is to have AI agents 386 00:28:37,200 --> 00:28:42,240 actually doing this autonomously for you. 387 00:28:42,240 --> 00:28:44,480 That is a very exciting project. 388 00:28:44,480 --> 00:28:52,180 Imagine AI building whole campaigns for you based on data and insights it has all autonomously 389 00:28:52,180 --> 00:28:57,040 gathered from your CRM and customer data. 390 00:28:57,040 --> 00:29:00,960 Being able to segment an audience and say, we think this is going to perform well and 391 00:29:00,960 --> 00:29:05,600 look I've created the landing page, I've created the AdWords copy, I've created the follow-up 392 00:29:05,600 --> 00:29:10,240 emails and the form and it's done it all for you. 393 00:29:10,240 --> 00:29:12,440 That is a real game changer. 394 00:29:12,440 --> 00:29:18,600 But I want to be clear, we're not there yet and who knows whether we'll ever actually 395 00:29:18,600 --> 00:29:19,720 get there. 396 00:29:19,720 --> 00:29:26,520 One of the tools I was most excited to see was the AI website builder. 397 00:29:26,520 --> 00:29:31,640 Many of us have spent hours, if not days fumbling around with different website designs and 398 00:29:31,640 --> 00:29:35,480 layouts and this tool aims to take that pain away. 399 00:29:35,480 --> 00:29:41,840 You describe your business, your preferences on style and voila, it will come up with a 400 00:29:41,840 --> 00:29:51,440 tailored site including layout, styles and pre-written content all through GPT 3.5. 401 00:29:51,440 --> 00:29:53,760 A few observations on this. 402 00:29:53,760 --> 00:29:58,720 It follows very much a wizard UI interface so you can put in some detail and there's 403 00:29:58,720 --> 00:30:03,380 a text box actually where you can describe your business and it's limited to 500 characters 404 00:30:03,380 --> 00:30:04,380 in your description. 405 00:30:04,380 --> 00:30:09,520 The more detail you put in there, the better the copy will be across your site. 406 00:30:09,520 --> 00:30:14,520 It seems like actually when you choose the style, what it's doing is it's picking a template 407 00:30:14,520 --> 00:30:23,320 and choosing some fonts and some colors based on that style but you're limited to say 12 408 00:30:23,320 --> 00:30:28,400 options and it might be like minimalist or fresh or these are the kind of terms that 409 00:30:28,400 --> 00:30:35,960 it uses and you just choose minimalist and then it will create a minimalist looking website. 410 00:30:35,960 --> 00:30:42,160 It pulls images from stock image libraries and adds them into the website and basically 411 00:30:42,160 --> 00:30:48,420 you get a done for you responsive website in seconds. 412 00:30:48,420 --> 00:30:54,360 What was really interesting was the response that this got on partner day. 413 00:30:54,360 --> 00:31:00,320 So lots of marketing agencies, design agencies in the room and when they launched this product, 414 00:31:00,320 --> 00:31:09,320 product yeah the response was somewhat muted shall we say. 415 00:31:09,320 --> 00:31:10,320 It's an interesting development. 416 00:31:10,320 --> 00:31:16,040 It's nothing like the Wix AI tool that has been previewed recently. 417 00:31:16,040 --> 00:31:20,560 That looks to be a whole different level but we haven't actually seen that in the wild 418 00:31:20,560 --> 00:31:25,680 yet so maybe that's just a style over substance until we get our hands on it. 419 00:31:25,680 --> 00:31:27,320 Next up was the AI chatbots. 420 00:31:27,320 --> 00:31:30,440 Now these aren't your run of the mill chatbots. 421 00:31:30,440 --> 00:31:36,160 We're talking natural language processing that can delve into your knowledge base, into 422 00:31:36,160 --> 00:31:41,800 the customer ticket histories and look at the different web product pages, your pricing 423 00:31:41,800 --> 00:31:49,400 pages all of that kind of stuff and use natural language to deliver personalized and quite 424 00:31:49,400 --> 00:31:53,040 nuanced customer support. 425 00:31:53,040 --> 00:31:57,360 You don't have to be techie to deploy this either. 426 00:31:57,360 --> 00:31:59,480 It's quite simple to get started. 427 00:31:59,480 --> 00:32:03,040 You make sure you've got the tracking code on the website and then you choose the assets 428 00:32:03,040 --> 00:32:07,760 that you want to add in to the chatbots knowledge effectively. 429 00:32:07,760 --> 00:32:13,720 So you would select all of the web URLs that you want to have included in its knowledge 430 00:32:13,720 --> 00:32:18,840 and then it will just start talking to people with that knowledge. 431 00:32:18,840 --> 00:32:21,600 There were a few other bits as well worth noting. 432 00:32:21,600 --> 00:32:28,760 They launched SMS marketing which is finally now baked into the HubSpot product. 433 00:32:28,760 --> 00:32:33,400 That's not necessarily AI powered but it's something that if you're a HubSpot user you 434 00:32:33,400 --> 00:32:36,720 probably found yourself wondering about. 435 00:32:36,720 --> 00:32:42,360 On the sales side the sales hub tools have had a complete redesign which is focused more 436 00:32:42,360 --> 00:32:45,720 on prospecting, deal management and forecasting. 437 00:32:45,720 --> 00:32:52,440 And AI is right at the core of that where AI is surfacing insights and trying to automate 438 00:32:52,440 --> 00:32:57,900 some of the grudgery associated with reporting. 439 00:32:57,900 --> 00:33:01,320 Actually that was one of the really neat demos that I saw. 440 00:33:01,320 --> 00:33:06,480 If you've ever tried to create a report in HubSpot it can be quite frustrating. 441 00:33:06,480 --> 00:33:11,700 What they've done is using prompts you can now find the right report and it will generate 442 00:33:11,700 --> 00:33:21,600 it for you so you can say show me the sources, marketing sources of our leads from the past 443 00:33:21,600 --> 00:33:27,360 month and it will create that report for you and you can add it to a dashboard. 444 00:33:27,360 --> 00:33:28,720 That's something that's been missing for a while. 445 00:33:28,720 --> 00:33:33,520 It's been a bit tricky to kind of get into. 446 00:33:33,520 --> 00:33:36,440 There are a bunch of other features as well. 447 00:33:36,440 --> 00:33:44,400 Things like customer support, so service hub, has an AI assistant that will summarize tickets 448 00:33:44,400 --> 00:33:46,000 and actually suggest responses. 449 00:33:46,000 --> 00:33:50,080 So if you've got a long customer query or maybe you've gone back and forth with a customer 450 00:33:50,080 --> 00:33:55,760 and had lots of conversation, if a new rep is jumping into that ticket to offer support 451 00:33:55,760 --> 00:34:00,840 they can see a snapshot of what the issue was there so they don't have to read reams 452 00:34:00,840 --> 00:34:02,840 of text. 453 00:34:02,840 --> 00:34:12,280 Overall, I think it's a really promising first step from HubSpot. 454 00:34:12,280 --> 00:34:18,880 Having seen the way that they've dealt the AI into the marketing hub in particular, it 455 00:34:18,880 --> 00:34:24,920 seems like they've taken two design philosophies. 456 00:34:24,920 --> 00:34:27,540 One is the wizard-based user interface. 457 00:34:27,540 --> 00:34:30,560 It will take you through a sequence of steps. 458 00:34:30,560 --> 00:34:34,800 It will ask you to describe something and then it will ask you which option do you want 459 00:34:34,800 --> 00:34:38,800 and then maybe give you another option with some variables at the end and then you'll 460 00:34:38,800 --> 00:34:41,720 hit submit and it will generate all of that. 461 00:34:41,720 --> 00:34:44,920 It will just take all of that information, put it into a prompt and spit something out 462 00:34:44,920 --> 00:34:45,920 for you. 463 00:34:45,920 --> 00:34:50,480 Then there's a more in-context editor. 464 00:34:50,480 --> 00:34:58,360 This is where if you're looking at the UI, so maybe you're creating a webpage and you're 465 00:34:58,360 --> 00:35:06,680 writing some copy or the text box where you edit the copy will have a button to start 466 00:35:06,680 --> 00:35:10,360 generating some content and it's a bit like Grammarly Go. 467 00:35:10,360 --> 00:35:17,660 If anyone has got the Grammarly Go Chrome plugin where you can use AI content creation 468 00:35:17,660 --> 00:35:23,440 anywhere in your browser, it's sort of like that. 469 00:35:23,440 --> 00:35:29,120 I'd say that HubSpot is laying some solid groundwork and the thing that's really exciting 470 00:35:29,120 --> 00:35:33,520 is when you speak to the product team and you can see that just on a personal level, 471 00:35:33,520 --> 00:35:43,480 they're excited about the possibility of AI agents being core to the backbone of the 472 00:35:43,480 --> 00:35:45,040 HubSpot platforms. 473 00:35:45,040 --> 00:35:52,160 Having played with these tools a bit, they're not reinventing the wheel or revolutionizing 474 00:35:52,160 --> 00:35:54,080 the thing. 475 00:35:54,080 --> 00:35:56,640 It's an evolution rather than revolution. 476 00:35:56,640 --> 00:36:04,040 The potential is what's grabbing everybody's attention and HubSpot's made some good inroads, 477 00:36:04,040 --> 00:36:11,840 but the future with AI agents potentially delivering entire campaigns and dealing with 478 00:36:11,840 --> 00:36:17,240 customer support tickets, that is where things look particularly promising. 479 00:36:17,240 --> 00:36:24,920 I've also had an opportunity to sit down with a few attendees at the conference as well. 480 00:36:24,920 --> 00:36:31,120 Here's a couple of snippets taking you through what a few of the delegates thought of the 481 00:36:31,120 --> 00:36:36,560 announcements and really trying to get an insight into how they're using generative 482 00:36:36,560 --> 00:36:40,720 AI in their marketing activities. 483 00:36:40,720 --> 00:36:47,840 I'm here with Tash Cartwright, Omnichannel Marketing Manager and Luke McNaney, Content 484 00:36:47,840 --> 00:36:54,640 Lead at Student Roost, the UK's leading purpose-built student accommodation provider. 485 00:36:54,640 --> 00:36:55,640 Is that right? 486 00:36:55,640 --> 00:36:59,640 We're one of the UK's leading, yeah, this is top three of us, but we are definitely 487 00:36:59,640 --> 00:37:00,640 in the top three. 488 00:37:00,640 --> 00:37:03,760 Don't want to overplay your role in the market there. 489 00:37:03,760 --> 00:37:10,960 You've just come out of Inbound 23 where the focus was all things AI. 490 00:37:10,960 --> 00:37:15,960 What are your thoughts on the conference? 491 00:37:15,960 --> 00:37:21,540 It's my second time at Inbound, so I came back in 2016, completely different experience 492 00:37:21,540 --> 00:37:26,720 with a few more years of marketing under my belt, attending slightly different talks. 493 00:37:26,720 --> 00:37:34,120 I learned a load about leadership styles, quite a bit about belonging, and then obviously 494 00:37:34,120 --> 00:37:39,160 the buzzword of the conference, which is AI, HubSpot's AI system specifically, but also 495 00:37:39,160 --> 00:37:43,120 how I could start to integrate AI into my day-to-day job. 496 00:37:43,120 --> 00:37:48,480 Yeah, AI was all pervasive right from the opening keynote. 497 00:37:48,480 --> 00:37:54,520 Luke, as Content Lead, this must be something that's been on your mind for a while, particularly 498 00:37:54,520 --> 00:37:59,720 since ChatGPT launched, is AI going to replace you? 499 00:37:59,720 --> 00:38:02,880 What are your thoughts now that you've been a bit more immersed in it? 500 00:38:02,880 --> 00:38:09,360 I think for me personally, it's been quite eye-opening and reassuring in a way, just 501 00:38:09,360 --> 00:38:15,200 because there is that skepticism there for my part and what that means for my job and 502 00:38:15,200 --> 00:38:16,200 my career in the future. 503 00:38:16,200 --> 00:38:21,560 I know it's something I've discussed with other people in the content world as well. 504 00:38:21,560 --> 00:38:27,560 So to actually go to some talks and learn about how these tools can be assistance and 505 00:38:27,560 --> 00:38:31,480 actually help you to refine your ideas and your content is something I'm really excited 506 00:38:31,480 --> 00:38:34,480 about going back to the UK and exploring in a lot more detail. 507 00:38:34,480 --> 00:38:41,500 Yeah, so we were talking off air and you were saying that you've not really thrown yourself 508 00:38:41,500 --> 00:38:44,560 into using any of these tools at this point, is that right? 509 00:38:44,560 --> 00:38:46,240 Yeah, not at all. 510 00:38:46,240 --> 00:38:48,480 So it is brand new to me. 511 00:38:48,480 --> 00:38:52,760 Obviously I'm aware of it and it's something that's been at the periphery, but I haven't 512 00:38:52,760 --> 00:38:55,120 actually had any sort of active use. 513 00:38:55,120 --> 00:38:59,880 I went to a couple of talks that were specifically revolving around AI. 514 00:38:59,880 --> 00:39:05,160 Obviously it was a bit of an overarching concept and something that was touched upon in a lot 515 00:39:05,160 --> 00:39:09,360 of the different talks, whatever the topic was, but some specific ones around AI and 516 00:39:09,360 --> 00:39:17,720 just how it can benefit and just enable anyone really writing content or delivering content 517 00:39:17,720 --> 00:39:22,800 in creating ideas and refining those ideas and just delivering in the same time as well. 518 00:39:22,800 --> 00:39:25,120 So it becoming a big resource. 519 00:39:25,120 --> 00:39:33,560 Yeah, I think that's the way to say it is as a tool rather than a replacement. 520 00:39:33,560 --> 00:39:40,400 In the Omnichannel game, was there any interesting developments in the HubSpot suite or any tools, 521 00:39:40,400 --> 00:39:46,880 any vendors that you spoke to at the conference doing anything interesting? 522 00:39:46,880 --> 00:39:51,880 Well it's a term that's been thrown around, again, more so in the most recent years. 523 00:39:51,880 --> 00:40:00,080 I think Neil Patel's talk is first talk about the future when everybody's using AI, how 524 00:40:00,080 --> 00:40:01,640 to stand out. 525 00:40:01,640 --> 00:40:05,360 He spoke about the Where's Waldo effect, which obviously we'd call the Where's Wally effect 526 00:40:05,360 --> 00:40:10,760 and that using AI tools could end up meaning that we all just look like we're wearing red 527 00:40:10,760 --> 00:40:12,640 and white outfits. 528 00:40:12,640 --> 00:40:18,160 However, he then lent into the fact that with Omnichannel, we're all trying to be absolutely 529 00:40:18,160 --> 00:40:23,520 everywhere at once and especially in the student space, it's that rule of seven, isn't it? 530 00:40:23,520 --> 00:40:26,360 You want people to have seen your brand seven times before they'll then come on and make 531 00:40:26,360 --> 00:40:31,560 a meaningful action, whether that's converting or whether that's just spending a little bit 532 00:40:31,560 --> 00:40:35,440 more time on their clicking about so we can collect some data on them. 533 00:40:35,440 --> 00:40:39,400 But he did say that it's a lot better for us to be concentrating on exactly where they 534 00:40:39,400 --> 00:40:46,560 are instead of trying to do loads and loads of different channels, maybe try a multi different 535 00:40:46,560 --> 00:40:48,240 content but on the same channels. 536 00:40:48,240 --> 00:40:52,400 So lessen your channels, invest more money into those channels, but then start layering 537 00:40:52,400 --> 00:40:57,160 your content on those channels so that people are seeing different pieces seven times, but 538 00:40:57,160 --> 00:40:59,040 not necessarily in seven different places. 539 00:40:59,040 --> 00:41:01,840 Yeah, that's a really interesting observation. 540 00:41:01,840 --> 00:41:07,800 I went to one talk with Neil Patel where he was talking about email marketing and he was 541 00:41:07,800 --> 00:41:12,680 talking about AB tests and hyper-personalized content. 542 00:41:12,680 --> 00:41:17,160 Actually, it wasn't AB test, it was multivariate and he was saying that they'd taken all of 543 00:41:17,160 --> 00:41:22,400 their customer data and created hyper-personalized emails. 544 00:41:22,400 --> 00:41:26,520 Every single recipient on their list got hyper-personalized and he said this was an email list of about 545 00:41:26,520 --> 00:41:31,320 1.5 million customers or subscribers, split it down the middle. 546 00:41:31,320 --> 00:41:39,440 Same people got the generic and 50%, so 750,000 people in each pool, 750,000 people got hyper-personalized 547 00:41:39,440 --> 00:41:44,480 and he said they run this experiment four times to see which one got the highest uplift 548 00:41:44,480 --> 00:41:47,440 and the highest conversions. 549 00:41:47,440 --> 00:41:51,880 He said hyper-personalization at that level, completely unnecessary, doesn't deliver the 550 00:41:51,880 --> 00:41:52,880 return on investment. 551 00:41:52,880 --> 00:41:57,400 She had personalization, standard personalization that we're already doing is fine. 552 00:41:57,400 --> 00:42:04,320 You don't need to put the time, effort and energy into trying to make it hyper-personalized. 553 00:42:04,320 --> 00:42:09,920 So forward looking then, going back into the office next week, what's the big takeaway? 554 00:42:09,920 --> 00:42:12,600 What are you taking back to student Roos with you? 555 00:42:12,600 --> 00:42:17,760 I will probably hold my hands up and say that I'm quite different to who I was seven years 556 00:42:17,760 --> 00:42:22,560 ago when I came to Inbound and that I feel like I may have become the kind of marketer 557 00:42:22,560 --> 00:42:25,840 that I didn't want to be, which is where I'm trying new channels and I'm trying new things 558 00:42:25,840 --> 00:42:27,800 in a safe space. 559 00:42:27,800 --> 00:42:31,760 I'm leaning into Twitch, I'm trying all these new cool channels that the students are using 560 00:42:31,760 --> 00:42:32,760 every single day. 561 00:42:32,760 --> 00:42:39,040 However, the idea of AI, I was leaning away from it and actually everything that I loved 562 00:42:39,040 --> 00:42:44,840 about my younger self and being more in my infancy in my career was that I just jumped 563 00:42:44,840 --> 00:42:47,760 into absolutely everything because I wanted to absorb it all. 564 00:42:47,760 --> 00:42:51,920 Stephen Bartlett from the diary of the CEO, he spoke about leaning in, especially when 565 00:42:51,920 --> 00:42:57,240 it comes to things like AI, which it's going to transform and it is transforming our jobs 566 00:42:57,240 --> 00:42:58,240 every single day. 567 00:42:58,240 --> 00:43:01,880 The next five years are going to look really different for marketers. 568 00:43:01,880 --> 00:43:05,580 If I continue to lean away from that, I'm going to get left behind and that's not where 569 00:43:05,580 --> 00:43:06,620 anybody needs to be. 570 00:43:06,620 --> 00:43:09,560 What we need to do is lean into it and use it in a way that's going to enhance what we 571 00:43:09,560 --> 00:43:13,720 want to do each day instead of actually going, okay, that looks scary and it looks going 572 00:43:13,720 --> 00:43:17,540 to take my job, so I'm going to pit myself against it. 573 00:43:17,540 --> 00:43:26,680 So I went to MIT on Monday before the conference started and I was walking around that AI exhibition 574 00:43:26,680 --> 00:43:28,840 going, nope, not a chance. 575 00:43:28,840 --> 00:43:33,800 Whereas now I've ended the week and I've actually used chat GBT to summarize a super long article 576 00:43:33,800 --> 00:43:38,060 about the new Google algorithm and send that over to my manager and I'm excited to go back 577 00:43:38,060 --> 00:43:39,440 and see how I can use that. 578 00:43:39,440 --> 00:43:43,640 That's a fantastic use case and lovely to see you diving into it. 579 00:43:43,640 --> 00:43:44,640 Luke, how about you? 580 00:43:44,640 --> 00:43:46,240 What are you taking back to the office? 581 00:43:46,240 --> 00:43:49,960 I think I also too will be taking the plunge. 582 00:43:49,960 --> 00:43:55,640 So I think just building on what Tash said, it's just a bit embracing the opportunity 583 00:43:55,640 --> 00:43:56,640 there. 584 00:43:56,640 --> 00:43:59,920 So as someone who's going back to the office next week to start and putting together the 585 00:43:59,920 --> 00:44:07,880 content strategy for the next student academic cycle, that I know full well that using AI 586 00:44:07,880 --> 00:44:13,880 tools is going to help me deliver on a lot of those outputs in a way that two weeks ago 587 00:44:13,880 --> 00:44:16,240 I would have really shied away from. 588 00:44:16,240 --> 00:44:19,280 So I think just that in and of itself is a big gain. 589 00:44:19,280 --> 00:44:20,280 Fantastic. 590 00:44:20,280 --> 00:44:23,000 So leaning in, not leaning out is the message there. 591 00:44:23,000 --> 00:44:29,840 Well thank you very much for your time today and I hope you join us on the show at a future 592 00:44:29,840 --> 00:44:34,460 date with updates about how you're using AI in your marketing. 593 00:44:34,460 --> 00:44:39,080 So I'm sat here with Joey from Tenshor. 594 00:44:39,080 --> 00:44:42,120 We're on day two of the event. 595 00:44:42,120 --> 00:44:46,520 Day one was all about AI, big product developments. 596 00:44:46,520 --> 00:44:49,240 What was your take on it, Joey? 597 00:44:49,240 --> 00:44:54,760 Yeah, honestly the first thing that I took away was HubSpot didn't take AI lightly. 598 00:44:54,760 --> 00:44:59,760 I mean they took an approach that was, hey we're going to use generative AI across our 599 00:44:59,760 --> 00:45:03,720 platform, sales hub, marketing hub, CS hub, all of the hubs. 600 00:45:03,720 --> 00:45:08,040 We're going to figure out how to help people be efficient and effective with our tooling. 601 00:45:08,040 --> 00:45:13,560 Really I think what it's doing is it's calling all the ISVs or the ecosystem that surround 602 00:45:13,560 --> 00:45:18,280 HubSpot to a higher standard. 603 00:45:18,280 --> 00:45:24,060 How are you guys going to leverage AI for your technology to keep up to the standard 604 00:45:24,060 --> 00:45:30,700 we're setting for making lives easier, giving people the freedom to focus on what they need 605 00:45:30,700 --> 00:45:34,040 to focus on rather than all the menial tasks that they have to do in between. 606 00:45:34,040 --> 00:45:37,200 Yeah, it's interesting you talk about the kind of community around it. 607 00:45:37,200 --> 00:45:42,080 As we were walking through the exhibition hall there we both observed that everyone's 608 00:45:42,080 --> 00:45:50,000 talking about gen AI, but in terms of actually having some deep integrations or interesting 609 00:45:50,000 --> 00:45:55,960 use cases, not a lot of it in the conference room floor there. 610 00:45:55,960 --> 00:45:58,160 Yeah, no that's a good point. 611 00:45:58,160 --> 00:46:01,720 I think it's because it's such a big topic everyone's talking about it. 612 00:46:01,720 --> 00:46:08,580 Everyone has an opinion as far as practical applications, people building smart solutions 613 00:46:08,580 --> 00:46:16,000 within their software innovating and creating customer centric customer first products. 614 00:46:16,000 --> 00:46:19,040 No one's doing that right now and I think it's for a couple of reasons. 615 00:46:19,040 --> 00:46:26,040 So the hype cycle, Forbes hype cycle puts generative AI at the peak of inflated expectations 616 00:46:26,040 --> 00:46:32,920 and that makes it bound for the trough of disillusionment. 617 00:46:32,920 --> 00:46:34,600 And so I think it's accelerated. 618 00:46:34,600 --> 00:46:39,080 I think it's going much faster than Forbes had anticipated. 619 00:46:39,080 --> 00:46:43,400 So we're quickly approaching trough of disillusionment, but what that means is people are expecting 620 00:46:43,400 --> 00:46:49,200 generative AI to do a lot and they're testing with it and trying to figure it out, but they're 621 00:46:49,200 --> 00:46:54,840 finding that there's a lot of roadblocks, a lot of things that generative AI is not 622 00:46:54,840 --> 00:47:00,160 capable of doing and at the core of it generative AI is just a small piece of the puzzle. 623 00:47:00,160 --> 00:47:11,360 It is not the peak of AI and it's not as broad as people expect it to be and so tied properly 624 00:47:11,360 --> 00:47:18,280 with business systems with other versions of like document AI, of vision AI, really 625 00:47:18,280 --> 00:47:27,600 utilizing a full like vertex AI platform, for example, like on the GCP, on GCP. 626 00:47:27,600 --> 00:47:31,200 It is, it's like a really small portion. 627 00:47:31,200 --> 00:47:36,440 We're running free generative AI workshops, little half day workshops tailored to companies 628 00:47:36,440 --> 00:47:38,600 to help people identify just that. 629 00:47:38,600 --> 00:47:43,800 Like how do you actually use generative AI in your business? 630 00:47:43,800 --> 00:47:49,000 What are some real practical use cases that you can build, implement and get rolling with 631 00:47:49,000 --> 00:47:57,400 today for software vendors trying to figure out how to use generative AI in their software 632 00:47:57,400 --> 00:48:04,120 offerings and for like manufacturers, retailers, people who are like in their lane trying to 633 00:48:04,120 --> 00:48:07,880 figure out how to adopt a new technology and make it work for them and where it makes sense 634 00:48:07,880 --> 00:48:08,880 and it's cheap. 635 00:48:08,880 --> 00:48:11,760 I mean generative AI is cheap to run. 636 00:48:11,760 --> 00:48:17,840 Like we're talking, if it's completely integrated in your business, every arm of your business 637 00:48:17,840 --> 00:48:20,360 is using it and using it really well. 638 00:48:20,360 --> 00:48:24,880 You're looking at a thousand, two thousand dollars of spend when it comes to like building 639 00:48:24,880 --> 00:48:31,840 your own generative AI solution using a Google or Amazon or Azure LLM. 640 00:48:31,840 --> 00:48:38,480 Yeah, so like do it, try it, keep working on it, iterate on it, like figure out a way 641 00:48:38,480 --> 00:48:42,320 to make generative AI work for you or talk to me and I can. 642 00:48:42,320 --> 00:48:45,720 Yeah, so that's very much what Tencho does, is that right? 643 00:48:45,720 --> 00:48:47,400 Yeah, so we're an IT consulting firm. 644 00:48:47,400 --> 00:48:50,640 We do cloud infrastructure like everyone, right? 645 00:48:50,640 --> 00:48:54,040 We can get you into the cloud, we can get you into a hybrid or multi-cloud. 646 00:48:54,040 --> 00:48:56,360 We think multi-cloud is amazing. 647 00:48:56,360 --> 00:49:01,520 It keeps people, it keeps cloud costs down, big cloud cost optimization play there. 648 00:49:01,520 --> 00:49:03,440 So moded me plays though. 649 00:49:03,440 --> 00:49:04,440 We do data. 650 00:49:04,440 --> 00:49:06,880 Data is our specialty place. 651 00:49:06,880 --> 00:49:11,640 So we do a lot with data from data lineage, data governance and security frameworks all 652 00:49:11,640 --> 00:49:17,240 the way to building a prod ready AI environment. 653 00:49:17,240 --> 00:49:22,260 And then we do applications, full stack software development, integration, modernization. 654 00:49:22,260 --> 00:49:26,920 We work with some really big names in the marketing industry. 655 00:49:26,920 --> 00:49:29,560 Specifically, we work with them to a MailChimp. 656 00:49:29,560 --> 00:49:33,660 But there's a lot of different applications that we're working on building, iterating 657 00:49:33,660 --> 00:49:38,960 on currently and all the talk is around AI, right? 658 00:49:38,960 --> 00:49:43,560 How do we introduce it into our culture and how do we introduce it into our platform? 659 00:49:43,560 --> 00:49:45,520 So yeah, that's what we do. 660 00:49:45,520 --> 00:49:46,520 Great. 661 00:49:46,520 --> 00:49:48,800 Well, it's been a pleasure talking to you and having you on Artificially Intelligent 662 00:49:48,800 --> 00:49:49,800 Marketing. 663 00:49:49,800 --> 00:49:50,800 Enjoy the rest of your conference. 664 00:49:50,800 --> 00:49:51,800 Awesome. 665 00:49:51,800 --> 00:49:52,800 Thank you. 666 00:49:52,800 --> 00:49:55,800 Thank you for listening to Artificially Intelligent Marketing. 667 00:49:55,800 --> 00:50:01,840 To stay on top of the latest trends, tips and tools in the world of marketing AI, be 668 00:50:01,840 --> 00:50:03,600 sure to subscribe. 669 00:50:03,600 --> 00:50:05,880 We look forward to seeing you again next week.