okay right it's 12 o'clock everybody and I want to be respectful of everyone's time for those of you on the watching the webinar replay um welcome this is all about AI for account managers the purpose really is to give you an overview of the general landscape of AI and we're gonna I'm gonna pass over shortly to Roy Murphy who's our resident emerging Tech expert to cover that for about 20 minutes then we're going to give you some tools for 20 minutes that are going to be relevant to your agency account management role and um then what we'd like you to do is to ask questions in the chat box and Alison is here to monitor the questions so we're going to leave enough time at the end to answer your questions and at the end of the uh of this webinar you will receive the recording so without further Ado I'm going to pass over to the lovely Roy Murphy to do a short intro and to talk us through the agenda thanks Roy thanks Jerry very short intro I'll make it super quick uh hello everyone just thinking that you know there's so many account managers on this call no one's doing any work are they agency land is literally bereft currently this is fantastic good to see good to see so many people though it's an exciting space obviously Lots going on uh loads to get through the next 15-20 minutes and then uh some brilliant tools from Jenny as well uh to give you some real context uh so yeah Roy Murphy that's me um please don't be disappointed when you meet me in real life my AI is much more handsome uh disappointingly um particularly as I'm not drinking this year and hasn't made my skin all my hair grow any better disappointing um so my founder of generate we are a new Ai and web free craft consultancy so right in the middle of his space uh fascinating stuff happening and we're working with our clients on AI strategy generative Sprints private development and all sorts of interesting stuff happy to chat at any point Roy at generate com so that's me uh coming up the next sort of uh 15-20 minutes a bit about the journals of landscape a bit about prompting uh some examples ahead of uh Jelly's brilliant tools and then we'll cover a bit of productivity and legal as well because the Landscapes moving so quickly and the legal landscape as well and the IP landscape move so quickly that um it's good to be on top of that as business professionals as account managers and to help our clients to kind of you know use this exponential Tech in the right and the most appropriate way so the landscape wow this is always a tricky one it's such a big and fast moving area that I've attempted to narrow it down to these three images now it isn't actually a boy band from the 90s gone wrong which someone said we were yesterday uh actually this is it so lots of things happening like you know chat gbt launched last year 30th November obviously um it's been happening before that but it really exploded and we've kind of seen over the last sort of eight months what um Mass adoption looks like what it means to expansion technology so on the left hand side I think maybe if you remember it we have a handsome looking fella in Balenciaga I.E the Pope I think of what happened around March was people started to realize that these tools were becoming really kind of good in terms of production and uh and that was one of those images that went viral and went oh wow this is interesting right well there's tools like mid Journey stable diffusion clip and so on so um so that kind of happened as one of those kind of big kind of uh touch points in the middle we've got Jeffrey Hinton the so-called Godfather of AI on the on the kind of on the the sort of doomsayer side and he came out in about a month and a half ago and said I'm out I've had enough I'm going to the beach this is far too much you don't know what you're doing you don't know what you've unleashed and I'm out I don't want any part of it on the other side of the coin and then our lovely image on the right is just what's happening this week uh so some really interesting moves and that is llama llama 2 which is Facebook who have launched their own uh or their second version of their llm which is a direct competition to uh to open Ai and GPT and that is a fascinating space and over the next six months we're going to see more and more of these things happening you probably have heard of loads of things around gbt and open Ai and the voice share that they've got and your clients are probably talking about but it isn't just about them so keep that in mind as we go through it really isn't just that and over the next couple of quarters it really will become apparent that that isn't the case and lots more of these large language models are in play from the big players and some others too so keep in mind last thing on this one is um disappointingly Hollywood's gone on strike so uh I can't watch any Netflix at the moment that's no good in new films part of the reason for them going on strike which is really fascinating and interesting for the future is the fact that the AI rights and the IP is part of that conversation so keep in mind that that often where Hollywood and other sectors like or work first other sectors follow so that's an interesting one to keep keep on top of um and don't go see the films because it can be hard to come come across over the next few months so um don't worry too much about what is in the press the Press is there to you know ultimately sell papers and and click some print we've been here before now you've got a balance of you know on the on the Optimus side and the doomsayers and so on but ultimately we've been in beforeton laundry right um in 1860 there was a warning about Ai and machine learning um in uh about 50 years ago there was typewriter licenses when typewriters come out because they were so scared that the written word was going to go away we've been here before admittedly not other level of AI but again you know we've seen these kind of um media panics and these media hives previously right it's it's just trying to keep a balance kind of view these things as professionals as account managers and we'll be talking to our clients you know because everyone's going oh my God what's going on with jobs going and this and that so let's keep in mind that balance is always always a good thing um just a couple of things on gbt in particular because that's where a lot of the data is currently you know we saw this a little while ago where it got to 100 million users in a couple of months and that hasn't happened before although very recently interestingly threads uh the new meta Twitter uh competitor uh did hit that pretty quickly and above that actually very quickly so it has happened since then but point being again Mass adoption exponential Tech everyone's talking about it your grandmother's probably mentioned it your dog uh your clients certainly have and and certainly interagency as well right so it's part of the conversation part designs guys very quickly which doesn't come across uh having very often which is why we're in the other GPT land which is general purpose technology and that is steam electricity the internet AI is part of that so we have to lean into it as professionals to advise our clients and to understand how our businesses work better um so how smart is it and again it's just API um open AR we're talking about and interestingly for a business called open AI they're very close now because it started off as a as a a non-profit and obviously we're not now and what we do know is you know the most recent and models dpt4 look at the the progress over 3.5 you know these models are massive um and these models are getting better by the way so in the background lots of things are happening going on but in essence um they are smart they like your AI assistant they're like your smart intern your um your a-level student without the attitude and I've got one of those so I know what I'm speaking about and so it's pretty smart but it isn't just about that uh it isn't just about open Ai and a few words aware uh open a are heavily backed by Microsoft so when we talk about open AI do keep in mind that we are talking big Tech um uh interestingly Microsoft have actually got a a stake and have come out in favor of um Facebook's llama as well interestingly so they're making some very interesting moves it matters to us as agencies and account managers though because our clients need to know about this so which of these tools which of these models are appropriate for us should we be using chat gbt to put any of our private data into the answer is obviously no that's a bad idea um but some of these models are much more enterprise-based cohere anthropic and there's loads of them coming out and you're going to hear more and more about these over the next few months particularly around the bottom two around meta and uh and Google as well so not just open Ai and when you're thinking about your clients as well as mentioned please don't put anything anything um that is sensitive into them but there are plenty of ways and we work on a lot with our clients of doing private llms which are based around security and making sure that data doesn't go anywhere else and isn't actually available to the public just in-house so um what can it do well I mean what can I do right doesn't mean to say it should but it can so research really really interesting um so loads of analysis that we've done on on our clients and our competitors and the market great for lead generation and tooling and putting those workflows together internally we're doing it as a new agency we're being lean and using the tools and eating our own dog food and doing that as well for Lead Gym uh great for nurturing kind of early in that space But again those ability for things like gbt to use and your own private models to use emails and plugins and use that workflow that chaining process in a more effective way great for summarizing and so on and can even help you with everyone's favorite timesheets but I'm going to move on from that because everyone's going to switch off and go for lunch um so a few things it can do Jenny's got some brilliant examples by the way so I won't bang out about this too long but um if you haven't um downloaded or if your business hasn't downloaded um or paid for chat GPT plus which is only 20 bucks so you really should um you won't have access at the moment to uh interpreter which is brilliant so this is data visualization so all these tools are coming out now which negate the fact that tools like GPT are fixed Transformers which means they are trained on Data before September 20 2021 in gpt's case and you're now able to use the web use plugins and use code interpreter to do things like that two brilliant small examples so put in just as a test showing the ftse companies give me the top 100 use that data into code interpreter and spat out these charts so there's no use of excel no use of bi nothing of that nature brilliant right and the clever thing here is the one that you get these wow moments of is when you see I didn't ask it for sector distribution I just asked it to do give me a line graph and a scatter graph for these things and it did it by itself and sometimes on these on these uh generative tools you go bloody hell that's I didn't ask for that but it's just being clever and these are getting cleverer on the right hand side something actually working I can mention briefly so we'll work in healthcare in particular and we're doing some compact analysis around what a onboarding process looks like for instance so we put some data in here's some examples not the real data and and the output you're seeing at the bottom again through interpreter so you can use these tools from text to interpret if you want to call it that or text Data visualization and that is going to help your jobs as account managers immeasurably in my view and it's helping us for sure so that is a a use case you could definitely take away which is okay if I've got a pitch coming up or a competitor I'm looking at something I want to bring my clients some new ideas blimey you know just go and use these it's going to give you 10 brilliant ideas visualize it for you and speed up your work so and you're going to look really really clever and probably get a promotion but don't tell them I said that right promising you probably all heard about it what is it well it's just basically it's the UI of AI so we put it um and this is this is why it's so big really and why that little text box in the form factor of Google 25 years down the line is so popular because it's so easy to use and you don't have to be a computer programmer uh or a data scientist to use it so prompting really important but when you're thinking about prompting please don't do this um by all means experiment that's really important and we've done plenty of obviously over the last year or so um but there's there's rabbit holes which you'll love doing and I'm into it I know Jenny's too getting into AI rabbit hole but you know it's great to experiment but have a process behind it because the outputs are directly related to the inputs basically so you can do these but don't go to that guy with a massage he's a bit of a doofus and he's making a huge mistake I made a huge mistake by spending three hours looking at his tick tocks you can do it but obviously it's not the only thing to do um what you can do instead it's something we follow we've got some downloads by the way um and we'll share those um is just use a process when you're doing these things what are you trying to get out of it like there's loads of ways of doing this and loads of information online but some of the ways we use it in the learning process is using some of these principles and just when you're using GPT or other tools just think about it what's the 80 20 rule why do I ask it that ask it to act like an expert try the high order questions metaphors and then still Manning you know doing the opposite of why won't this work why won't this SEO campaign I've thought about all this uh piece of technology or analysis work to do those processes we've got some um some examples of those in our download uh happy to share that with you guys as well which is kind of useful um and yeah they're there three things out of that one really there's a brilliant course uh that we have no affiliation with it's just fantastic we send link to and that is by open AI but don't worry about that it's a little bit heavy on the development side but you don't need to do development or understand it do it so for account managers perfect and you know what it will just teach you probably in an hour I think it's a 90 minute course uh over 90 minutes it'll tell you 50 times more than a million of those tick tocks will because you'll get to understand how the thing works what inferences what future and one shot learning are it's fantastic so Carl said um too much about how good it is and we've got some little cheat sheets so we put together for different things and our process so happy to share those uh to share the love in terms of generative because it's such a brilliant space we'd like more people to be involved um look you know not to ban our drum we're in this we're in the space and we're working on all of these things this is brilliant I won't talk about too much we're working in ecoms we're doing content and production work and visit attractions campaigns in finance and across the board in loads of different um uh really interesting generative ways with our clients and summarization social and then further down the kind of more development route um using kind of it's very interestingly using company data as I mentioned earlier about Fiverr data and using the fine tuning of um of GPT and other models to make those models work for our clients so it's hell within their business and they can use their data in a much cleverer way and that is really making it machine readable turn it into vectors and then outputting in the most appropriate way I say loads of that stuff going on fascinating stuff and great for us as agencies um and those are tools like experiment these are all things that were done text to text to um straight into GPT um you know there's business canvases we've made I think we've done about 30 or so web apps and then use some of those apps into our workflow which again you guys can do and trackers timesheets expenses um um all sorts of different ways of applying workflow and tooling into our job and again as a new business we're using this quite significantly and talking about on our podcast which is how do these things work and how do they kind of come across are they production ready and they're not always but I think a lot of it is is really is really useful stuff so keep an eye on those um right just a couple of brand examples there's tons of these out there but a couple of interesting ones and again the first half of the year we saw a lot of PR around this and in we're definitely from our conversations um and our account teams and so on with the clients we're talking to it's much more around okay we understand it now we know we need to do something but what should that be what's the use case what's the guard rails what's the safety how does the IP work are we going to get sued um what's the reputational damage all these questions are in play and what you've seen in the first part of the year is Brands jumping in and being having guard rails over their work so you know simple things like Bex creating AI beer I bet it tastes absolutely disgusting I wouldn't touch it not this year anyway um and uh doing a PR campaign and packaging all AO based nice little PR story you know did they really do it then I'm sure we did Coke you know in the middle gray brand straight in the middle of this created a platform whether they've got a partnership with Bain and some of the bigger agencies and they've gone and they've gone all in on this um did a brilliant campaign in kind of March April around using their assets and being generative using influencers and that's really cool and they've done a whole bunch of other stuff including some adverts as well a large Brands like um Adidas are tying up customer personalization with generative which is really interesting and we're working in that space as well within the web free side and the transparency so that is personal lots of much creative but personalization within generative to affect your journey through their um through their portal to sell you more products to sell better more personalized products fantastic case study so lots of those interesting things we're going to see more and more of those as you go through the next couple of quarters uh right so forecast for productivity so what's what's the business side of this saying what's it all about so a couple of top level things we've seen and I should expect you know the larger consultancies all over this the Baines and the mckinseys and the Delights and the pwc's all brilliant all brilliant businesses not as fast as us obviously but brilliant nonetheless um but from the top down they've done a brilliant report actually really good you know and this has been they've been widely shared you know up to 4.4 trillion GDP in the next 10 years equivalent to you know 3.1 is the UK in 2021 it's probably a bit less this year and so there's big numbers in here and big plays uh coming in coming into play and their use cases what how they're seeing it and we kind of I can concur With It software engineering customer care or service product r d sales and marketing so we're seeing more in the sales and marketing and customer element but definitely in software engineering in terms of productivity with co-piloting very big so they're the ones that they see and interestingly 60 is worth 60 or 70 of work time could be automated now as agency these we're all about efficiency right um how much busy work do we do on a day-to-day basis probably quite a lot right and so there's productivity gains they're huge ones actually for us and for our clients so um so yeah that's why that's why this is uh I'm so bullish on this um um apologies for that image that was a last minute Edition for some very bored looking people uh on this Deloitte on these Delights that's 52 of heard of Jenna in the UK of the 5 000 sample I'm not sure that's right 80 are using it for work interesting okay so that's like uh four million people and this is the interesting one a roughly 40 assume that it's correct or believe it's unbiased wow it's not the case it is the case depending on how you fine-tune your data but actually interestingly how that would that will then shape over the next few months of customer confidence and how that shapes the aid our clients and their services with using generative and that's conversations we are definitely having right now so this is the snapshot today by the way might be different in six months um a couple things that we've seen not ours but another study we saw I think was from digidate of 100 agencies who was using it as back in I think May so may have changed 70 of agencies I'm not sure seven you're actually using it or talking about it but who knows uh Brands up to 50 is probably up a bit higher at the moment and from the same study of a roughly 100 agencies and Brands um these are the kind of core areas and I would that would correlate with our with what we see as well so research and copy and content definitely because they're kind of easier to get to grips with you can see it as an output and marketing is part of that for sure um video audio is a nascent area we're spending quite a bit of time in actually because text to video isn't quite there we have some clients actually working we're working on the moment on some on some clients in that space it's not quite their production wise but it's getting very there very quickly audio as well not to forget about that important point the data management side on the bigger pitch a little bit harder because that's typically into private data and how you're going to incorporate that as part of your transformation as part of your workflow and so on so that is typically a little bit hard um a little bit harder to to get uh to not get started but to do um but it's getting there more and more the conversations we're having and I'm sure you guys will be having with your clients if indeed in this space um are certainly moving down that route very quickly um okay a couple of other ones just on this um great study come out and in customer service and the key things here were for us as agencies is interesting is that the problem with large organizations and sometimes more ones is um you know more than you tell you can't get that information out to people so in customer service for instance there's uh there's a great study where um GPT and generative were used alongside customer service agents and it increased their it rocketed up their their um their productivity actually and they're resolving and the customer happiness as well but a key thing for businesses and our clients particularly and maybe some of the largest agencies is that in the middle so the low performers were skilled up very very quickly in this particular use case in two months instead of six months and the high performers did work as they would anyway because they're high performing they share their knowledge they just um they integrate with new technology they're not scared and they're they're you're high performers it was the middle layer where the productivity gains and the challenges were so as businesses advising our businesses um as accountant business people and that's potentially where some of the large numbers you've seen thrown around by Goldman Sachs 300 000 people uh BT I think saying 10 000 people those middle jobs are the ones that are going to be at risk because if you're low lower level um low performing staff should say are able to um so are able to uh to perform better quicker using technology you're going to have a ShakeOut and I had a brilliant conversation with a very large organization last week about the the dichotomy the push and pull between the c-suite rubbing their hands is a bit strong put together saying this is brilliant productivity Roi we're gonna we're gonna rationalize it's brilliant we're going to make more profit to the human cost of people saying but that's jobs and there's people I'm scared of my job so there's a push and pull there which can't be solved today but it's a it's a direction of travel um on the from the very bottom up from our side up rather than top down um this is what we're seeing we've got um a couple of these but one one in particular a a a significant um six-figure um climb uh with a 60 direct cost saving using generative and that is an absolute Rock Solid fact because we've just done it so it and we're doing more of these so it really is that is an absolute fact so this stuff does work internally 20 to 30 I think with development in terms of co-piloting and other tools um uh for our development stuff and it ranges it depends what you're talking about 15 to 75 productivity I did something very small yesterday which was 100 productivity didn't have to do it and I did it in interpreter 15 on some of the tools and again being lean as a business Little Agency we're using a few of these tools to chain our workflow together and it's certainly 15 and upwards from there so all to play for basically um lean into it because this stuff really does work um right a couple of things quickly on the uh the legal landscape um this we could do a whole you know whole day on this really but where it sits at the moment is obviously regulation is is behind because this has moved so quickly and we don't see this very often which is this amount of technology and this amount of speed so we've condensed maybe five years of innovation into six months so no wonder is it but everyone's getting their activated very quickly there's a political social behavioral technological reason for that we're in general purpose technology land here which could mean huge amounts of gains for um for Sovereign AI for the UK for EU and America China et cetera different acts across different things EU Act is happening probably more on the gdpr kind of level labeling and safety levels and laws will come into play the UK has taken unusually not being political a very good view I think um of pro Innovation and decent centralizing at least for now what the regulations will be in terms of and that Pro Innovation approach for for generative and AI China interestingly are in I think Apple named cyberspace Administration committee I think is um is coming on the 15th which will be the first one globally I believe um which no doubt will have some correlation with um how the country is run but that'd be interesting of course the US become you know big area very Pro Innovation but lots of interesting things happening there around copyright and IP in the Senate and the um uh the FTC as well with openai just kind of lots of pain of them but are talking about what does it mean so they're taking it seriously which is brilliant and then there's lots of things happening around is AI creative the Grammys come out I've got a week ago saying you know um we can accept AI creativity for next year but it has to have a human component I think we're gonna see more and more of these things and then Getty obviously is a creative business are also doing a lot of um work around um suing people uh because someone over there might have been trained and but I don't be interestingly uh have taken a different view so last couple on this um I've been copyright just on the creative side and again I know it's not all creative agencies but just to understand where this is going for the tools that we use every day um Adobe and Charter stock have both uh identifying um Enterprise customers if they use generative tools and their generative um imagery to create they will they will pay your legal costs which which well I heard it I know my chair I was like wow that's amazing now I'm not a lawyer a car so that's definitely how it's going to shake out but this is the direction of travel so you've got all the incumbents over here and it's good for us a nice businesses but the incumbents are over here saying wow this stuff's moving quickly we can't miss this boat so so what's happening is the trusted one the shot of Foxy adobe's the Microsoft is yours the aws's alarms Etc are doing exactly that which is trust us you trusted us already you need to trust us now so that's where this is going very quickly so that's fascinating and that that is really a space to watch and last couple then very quickly uh uh AI doesn't always go right it sometimes goes wrong um there are ways to game the system it's getting better uh that this is one of our favorites on the left which is asking AI about downloading power websites it says absolutely not I can't give you that ask AI well which one should I avoid so I never go near them and it shows you it's kind of funny um not that but I've got pirating films obviously oh that's tightened up recently actually so I'm not sure it's quite as easy and hands are just hard to do in AI from the creative side kind of funny and because there's so many joints it's the hardest thing and you know what in the Renaissance they couldn't get hands rights either and if you look at cage dwelling drawings also similar and don't do this one Whatever you do the Avianca example which is a lawyer in Latin America who um used chat GPT to make up some case law got found out and I think he's in jail currently so please don't do that account managers and agency execs as a terrible idea right uh last bit then if you want a bit of a laugh slash some truth and uh the definitive generative AI guide for account managers specifically for you brilliant people who have come on this call for no one else go to generate.com account book as a brand new heart of the presses AI guide just for you which will give you a giggle over your lunch break so go and download that because it's fun and uh it's only 50 generated because I'm a brilliant copywriter obviously so many of me right that's it from me uh three things really look you know it's a brilliant space we're in general purpose technology land you have to lean into it even if you don't believe it you don't believe the trust is there you have to understand it because your clients are going to be asking about it and you need to be productive and you need to have even answer it the answer is no because of X that's a perfectly good answer but you need to have an answer and have an understanding of it so be AI enabled doesn't mean AI is everything and educate yourself loads of resources out there it's not difficult and don't forget very importantly that we're early right this is really early in you know not too late you can get involved get yourselves of your own careers and so on for your agencies and definitely for your clients so happy to chat ask questions looking forward to Jenny's brilliant piece on all the great tools for you guys as account managers and give me a shout at royalgenerate.com Roy thank you so much that was so informative and so um thorough so thank you I'm just going to share my screen and just um while I'm doing that uh just to let you know that I've actually invited Sharon Turek who's a U.S uh lawyer specializing for agencies and is talking at the upcoming AI for marketers Summit meeting so I'm really lucky to have her as a guest on the podcast so look out for that if you're interested in the legal side particularly so look I'm going to share some tools that you could be using if you're not already that I've divided them into four categories and as Roy said they're mostly about speed and productivity because I wanted you to come away with some specific tools that you could use in your job right now so the category one is Service delivery so you have to be good at getting stuff done in the agency and organizing yourself so AI can help us organize our work load and also speed up our self-development so let me show you a couple of uh tools if you're not already using it todoist with an AI assistant is fantastic you can prioritize your tasks as you enter them and if you follow the eyes and how a matrix method which is the labels down on the bottom right hand corner you can actually label the tasks as you put them into your todoist list it automatically breaks those tasks down into steps you can use the tool to time block and it also integrates with your calendar system and is available on desktop or mobile so if you're kind if you're the kind of person maybe you're just starting out in an agency you have to organize yourself this is a great tool to use the second one is glass glass but again you might be familiar with already but I think this is really going to help self-development so you've probably currently got lots of different resources that you use to learn about your job so let's suppose you're an account manager in Performance Marketing you may be reading the drum you may be looking at podcast uh listening to podcasts watching YouTube videos to accelerate your learning you probably are using learndigital.com with Google and I put the link there I've just read it out but it's called learn with Google but they do free digital strategy um courses what this tool will allow you to do is basically consolidate all of your learnings into one place in one comprehensive dashboard so I've just got a screenshot there of what it looks like when you hit the drum if you pick an article that's resonating with you that you're learning from you can highlight the areas that you're learning from capture it um in glasp and that will mean that every all your learnings are in one place here's an example of how you use it on YouTube so any YouTube video you imagine you don't have to sit through hours of a YouTube video here's me interviewing Roy on our latest podcast podcast episode if you have glasp which is just a Chrome extension this comes up you click glasp and then you it summarize the whole YouTube video just into one succinct um paragraph in Jack gbt so use that and I think that's really going to save you lots and lots of time reading Reams and reams of stuff on the internet the second category is about client experience and I think particularly AI can help us enhance our client relationships so let me share a few tools the first one I've been banging on about called Crystal nose and this is an AI driven app that plugs into LinkedIn so let's suppose you're looking at your client's profile on LinkedIn they happen to be quite active this will tell you what kind of communication style they have it would also help you with the do's and don'ts of how to build rapport with that person so you could be researching your existing clients but also if you're pitching for new business you could be looking at your prospects on LinkedIn and this will just help you build that Rapport and communicate with them in the most appropriate way that suits them the next one is poised again I had the founder of poised on my podcast episode but this is like having your own private communication skills coach so it plugs into any virtual meeting platform it will tell you if you're talking too fast like I am now if you're using hedging words like sort of kind of Etc it will just help you sharpen up your communication skills it will tell you if your energy is low if you're asking if you're not asking enough questions if you are not pausing enough so you can either have this live box on the screen telling you real time but you but you could close that off and also have a report afterwards so you can track your progress um over time fantastic tool I highly recommend it the next one is um word tune spices and I want to give a shout out to Conrad Sanders who came up with this tool um on my recent chat with him on the podcast this is also like you'll have a private writing skills coach that's going to get you better at writing emails proposals pitch decks you know how many times I I know we're using Gmail for example and it auto corrects but this this tool will help you add things like statistical facts or analogies or expanding on a point that you've made so it's a really comprehensive tool it's worth exploring to make you better at writing fireflies.ai now again you may be using a transcript transcription tool with your virtual meetings with clients but the reason I'm sharing it I used to use otter.ai but I'm now using fireflies specifically because it integrates with so many different um project management tools but also it gives you a really succinct summary and action points so again saves you tons of time you don't have to try and remember what you talked about in the meeting and start writing a big contact report it's going to make you look efficient when you send a really succinct summary with the actions directly after the meeting so check that one out if you haven't already obviously ask for permission from your client to record the call but it also gives you a voice of the customer so if you imagine that you're capturing all of this data from your client conversations you can then pass that information onto your marketing department so that they they can create they can integrate that voice of the customer into them into your agency's marketing efforts third category consultancy so this level is probably for more experienced account managers who are expected to understand the client's business to come up with relevant ideas and to really immerse yourself so this tool can help you speed up with client research because if you're handed a new uh business you know we've just won this account yay now it's down to you you want to get up to speak quickly on on what's going on in that client's business and in the past you have to break through lots of newspaper articles and Industry press and looking at the websites use this tool immediately it's Roy mentioned it it's uh chat gbt4 code interpreter it's the 20 pound a month you go into settings you go into um it's in beta mode at the moment as Roy said it's just come out but you can basically upload any piece of data so I'm going to show you an example if you were to download for example the Chairman's statement or the company's annual report or or the transcript from the investor relations meeting um it you can just it will summarize it for you so you can give it a prompt I've just given you an example I mean I'm not I'm not great at prompt but I just said like take the following data and create a visual graph that's it so this is what it gave me so let's suppose you've said you've just won the Diageo account and you're on the Smirnoff brand you're like where do I start so you go to the Chairman's statement and you think well that's lots of copy and lots of text you go to chat GPT code interpreter and you just say Give me a summary of this and make it visual because I haven't got time to rake through all of this look what it gave me honestly I couldn't um I could just couldn't believe it you know I I started off in the dinosaur years of account management where everything took a million years here it is in one so you imagine you're just about to go to a client meeting you've only got two days because everything happens at this this you know speed of knots in an agency you've got no time this tool can help you with your learnings last category number four before we go to um your questions this um this is a way of finding out and and identifying some future growth opportunities in the account so again if you are a senior account manager account director of Client Services director it's our job to look at the forecast and and to come up with relevant ideas to present to our clients so what can you use for that again step one use whatever data you can download so again if you if your client in the in the UK is a limited company you can download their reporting accounts from companies house similarly if it's a PLC you can go to some tool like a website sorry called Seeking Alpha and you can download the transcript from the quarterly investor relations meeting again it's reams of data which you want to quickly summarize and get the gist of step two is to I ask it to identify ideas so I've put a couple of tools in here booklet is one and the other one is chat GPT prompt splitter because what I've found through testing this is the bigger the the PDF that you upload sometimes it struggles with a huge size so what you can do with these two tools is actually select the piece of the data that you're interested in summarizing and analyzing it then give it to jatch EBT and ask it to visualize so again The Prompt nothing fancy tell it what who you who they are you are an account manager in a digital marketing agency please identify the opportunities that I can provide in terms of services based on this data that I've just downloaded from the client that's it so um here's an example Dobby's garden centers so I went to companies house I downloaded the document it was 49 pages um there's a there's from seven page seven to page nine it says future developments and just from that information from a few pages I said come up with some ideas that I could present that is relevant to where they're going as a business and I was blown away I've been so excited to share this it's just phenomenal it came up with nine opportunities these are nine digital marketing ideas that I could go back to the client with or it could be just stimulating my thoughts on the kinds of ideas that we could present to the client and Behind These nine opportunities it wasn't just one line that gives you context so I just had to pull out one because there was just too much information but look at this one of those categories they said was local marketing for new and redeveloped centers so turns out dobbies dobbies is um Garden Center if you don't if you're not familiar with Dobby's um they're saying like you can as the agency you can help in the promotion of the grand openings and the redevelopments through local SEO event marketing and Community engagement I mean how cool is that even this if this is just the start of the 10 it's certainly going to put you ahead of the head of the game and it's going to start stimulating some thoughts for where this could take you and then finally using the same tool code interpreter on chat gbt4 let's suppose again your senior you give them an Excel spreadsheet so I made this up it's got four tabs and I said okay here's all The Client List here's what we build for the last three years and here's our 2023 forecast visualize it for me just with that data and also tell me from 2022 if I've got a 10 growth Target tell me what that growth Target would be based on the current forecast as of July and again I just was basically so bloody excited about this um so yeah they're they're the steps and this is what it gave me I mean I could probably even go back and say make it look a bit prettier but you can see that it's laid it all out um and it's given me like your 10 growth Target from 2022 would be 1.92 million so I'm just thinking about for the account manager if you want to look at the client spend for the last three years ask your Finance director for any spreadsheets they can give you put it in private mode on your chat gp24 tool upload the data and ask it to analyze for it for you because you're going to get client patterns of spend you can even break it down by types of services and this is like having your kind of own data analyst by your side during your working day so that's where I'm going to stop we haven't done badly on time actually Roy it's 12 40. I really hope that's been useful but what we'd love to do is I'm going to stop sharing and um ask for questions so Alison I don't know if you've had the opportunity to or is anyone asked any questions so we've always very kindly answered a question in the background whilst you've been chatting um and actually we've just had our first new question pop up so Joe has asked when dealing with financial data um it's really important if we're using AI tools to help with this it has to be super accurate I would hate to quote any incorrect figures when talking to a client oh my God I I agree with you Joan is that Joan Palmer Joe Greene no sorry Joe Greene hi Joe um Roy you probably want to answer that but I I've got my view because yes you have to double check but Roy you've probably got a much more comprehensive answer exactly right you've gotta you've got to see some of these systems as like I said earlier the intern it's you wouldn't let them run crazy with your with your clients so in some cases it's better to use that privately I said private mode or or preferably in your own private um uh llm or fine-tuner then um and double check it now of course you may as well I was is that actually helping but it is because the the predictive and generative side of it is giving you potentially other routes to travel in relation to what you do with that data so so you have to check it's only only sensible checks and balances particularly when it comes to Core Business activities that you know a lot of things there by the way but again because it's so new and because the direction of travel is and there is actually some stuff happening anyway around more private data and financial data those use cases and those smaller uh specifically fine-tuned models are actually already coming out anyway so in the next few months you're going to see more and more of those which are tuned directly into um uh Finance or operations or or um or Healthcare and so on and of course there is you know there's obviously lots of the large businesses the sales forces and so on are are incorporating those tools into their process anyway so you're probably going to find that some of those things are you don't have to go and roll your own that's not right for everyone either so some of these larger organizations and the tools you're probably already using if they're not already doing it they will be pretty soon so so keep an eye on it then I just add to that Roy because I agree it's a great point that you've just made is that look at look towards your existing Tech stack first like what are you using currently with your agency because AI is probably if not already having AI capabilities built in um anything in the public domain obviously you can um analyze quite freely Joe um it's just the stuff that is going to come out up with from scratch that you need to double check I suppose and Roy a question for you when you said either put it into private mode or use your own llm do you know of any agency currently that is using their own llm maybe the big networks are because I I know that what you referred to before but anyone any other agencies on an independent basis are using their own llm and we're well we're working on ours obviously you are yeah yeah so I can't I can't speak for other agencies um uh the conversations we're having and again with partners and obviously we work with agencies too by the way just a point to make um as as clients as well and of helping out some agencies so yeah the conversations are in that space But I think it's quite early still some some people are are bullish on it and say well we're going to use this and some people are not I think like I say it's fair to say the larger networks and holding companies are all over it the wpps of the world and so on and because they've already they've already done that basically and I think even publishes uh did launch an early version of this I think it was uh was it called Claude and that was an early version of a couple years ago and what's interesting what kind of laugh that which there they made a big thing of he can actually over the last um uh the lion Festival saying we told you we were correct three four years ago when you were laughter are hugely multi-million dollar AI um platform so the large ones definitely and but I think if companies aren't and there's maybe a SAS opportunity but um uh if companies aren't doing it it's it's gonna happen pretty soon so like Roy you work with agencies as well as kind of marketing clients Tech clients can you help navigate that space like if there's any agency listening that thinks that's a really interesting thing that I'd like to explore is that something that you help them with yeah we're doing that right now actually uh on some really interesting use cases around that I've mentioned earlier like the fine tuning um so that is um exactly that so the couple of use cases that we're looking at um there is um uh new starter onboarding we're looking at that um there is interestingly financial data in relation to internal companies that we're looking at as well uh so we are doing that so yeah happy to happy to have a chat um it's fast moving the tools that we're using now are also being updated and new a new um incumbents into the market on a daily basis I mean we it's interesting we stopped um we started in December actually uh uh putting together a updated tool set of different types of tools of course we keep on top of it but we stopped uh when it hit over a thousand we stopped and said this we need to take a different view here it is actually very difficult as a growing fast moving and growing business as well to keep on top of all the tools the right approach of course is what you use as with anything and again not not to not to forget that you know technology's a catalyst it's not everything even AI still needs human input and before it's sentient and it isn't currently um uh but super intelligence within a decade says Sam Auckland CEO of open AI who knows and AGI is on the way you know um uh baby AGI and agents and uh Auto GPT all these other things you may have heard or not are coming potentially not not production ready yet so it's all about use cases these are technology tools as a catalyst right we just have to remember that as well yeah start with the tasks that you're doing analyze the task break it down into steps and think can I get um any tool to help me with that so I agree Roy Allison have we got any more questions yes see quite a lot of now coming so actually one of one of them that's uh linked to what you've just said is um do either of you have a specific um model that you would recommend to use to use AI on mundane tasks like click reports because an anonymous attendee is asked or mentioned that they spend a lot of billable time on this and they want to put pressure on their vendors to help them so any model um that would be better in AI to use on click reports but what was the report sorry click reports okay um yeah well obviously the um quite recently um and again we're into API land here as well because you know you don't want to be putting anything that's private out in the wild um but anthropic um so claude's claw two uh which was launched I think a couple of weeks ago which has a a much larger context window is interesting so GPT I think is eight to ten thousand if I'm wrong I might be wrong but Claude has a hundred thousand so the ability for it to remember larger type reports is really interesting it's cloud.cloud.ai so that's anthropic one of the big one of the bigger kind of more independent plays although are in bed with them to the large Tech as well and would be worth a look because certainly in API level then you can use it privately use it within your Azure whatever whatever system you're using use your productivity if it's probably available fine but that's a it's a useful tool um uh it does hallucinate still a little bit towards the end and that's what happens when the context spinner runs out and the memory runs out but I would probably recommend that if you're going to try and compare something uh it might be useful great thank you um a couple of questions that sort of merge into each other um one is um from Pippa McGuire about um how to give confidence um to clients that um if you're using um AI in your agency to help with work on their projects um especially if they're the clients companies haven't actually given them guidance yet or blanketly told them either to use it or how to use it and also um alongside that um how do you make sure that you're using eye to create something personalized and bespoke is there a danger that the same sort of information will be will be generated by different people in different agencies first part um hire us to do an AI audit we're doing loads of them being slightly glib but that is what we're doing which is a great question though to be honest because what we're finding and this is this is literally changed in the last probably two months um in also is this the the conversations up until around March or April were uh what's this in there yeah we use it with a literally met in May onwards it's been okay how are we going to use this so we need to take a step back a little bit what's the reputational risk what's the security risk uh what's the risk um to our internal staff and what's that what's the ROI and that is a a degenerative AI strategy Sprints that we are doing with clients in exactly that format because you're absolutely right it's okay to test something and you know you're and even if you haven't mostly most businesses don't have a a generative policy should your staff be allowed on company premises so therefore as part of your I.T to use chat GP there's some businesses you know we know some businesses that haven't they say no we're just going to not do that because we don't want people going oh just I can't be bothered uh writing this brief for said client so therefore No One's Gonna know if I put it in I put it on private mode we don't really know where that sits really to be brutally Frank um if you're doing that there's a risk to that but again and people are fundamentally I wouldn't say lazy but this is me by the way you know I right it's a tool I can use and it's appropriate I'm going to use it so I think I think AI strategy becomes table Stakes to be honest uh that's that's the the key answer because testing and POC is great too and we have done that as well by the way you don't have to go and do that as long as you've got a particular use case so let's keep the the social part of it or let's look at our content production and let's look at our summarization or read my workflow we've done all those things but it really depends on on the use case so um definitely have a strategy and what's the second part of the question sorry uh from Lisa Chan um as more and more people start using AI how do we ensure we get a personalized and bespoke service or output yeah so uh for the interesting thing of course because it's generative or predictive you don't get the same result each time which is also as a boon and a slight challenge right because if you if you need to get and this is this is again the conversations we're having continuously are but if the output is changing hell don't know which output is correct great question and we'd certainly within the visual media um part of generative so um and what I mean by that is that the images the text of video the uh and the tools being mid-journey stable diffusion Dolly and those kind of groupies of um and Firefly Adobe as well they've all got different ways of doing weights they've all got different prompts and if you're if you're um uh kind of experiencing off in understanding how that engineering part Works visually you can see these images you can use them I've done it with some quite big Brands actually um which is you need to have you know in that simple to understand where you need to have the uh the brand Essence correct right you need your tones and everything else it needs to be kind of in a certain way but we've done it with it with a couple and there are ways around that definitely and and other things you know again it's context isn't it what are we trying to do here if it's Finance it just needs to be correct it can't hallucinate therefore smaller private large language models fine tune to your own purposes are the way to go 100 um so yeah so I guess I guess it's um it's an open question though isn't it though um you the ability for generative is also some of its challenges which is okay but no no I don't want it to do more creative things I just want 10 of these things so that is that at the moment is uh is needs a bit of wrestling about wrangling to be honest um but you can do it just need to be experiencing it then I add a couple of points to what you just said Roy I would reiterate the point that you've said about having a policy um I've been following a lot of what procurement are talking about currently and they're being briefed to ask their agency what's your AI policy so I've I've picked out a couple of articles that were saying the same thing so yeah there is a bit of a demand for that recently and I was speaking to an account manager two days ago who'd been asked by her client to provide a prompt that would enable them to replicate the work that they had done and it was a series of headlines so that is it really is important for us as agencies to have a policy the second point I just wanted to mention was a tool that um is called writer.com where you can upload your brand guidelines and have that tool come up with um copy that is in line with your brand tone of voice so I just wanted to add a couple of things there and there are actually a couple of questions that lead into that um about the the clients potentially taking um work away from the agency [Music] um one of them is what do you think about AI being used to transition into selling value instead of selling time and they've asked you Roy do you still charge for work based on hours and then also this danger for agencies that in the mid to long term that Brands marketing teams begin to bring more and more work in-house obviously it's about future proofing but how would you respond to that yeah it's a great question um no we don't is this your answer sell on to on time someone value um because in a world where anything's possible and generative is part of that and certainly as agencies are more the more creative agencies and SEO and performance ones too it can help across the board so I think that the um the answer really is well first of all no we don't we we do on on them on value but um the it's it's quick it's fast moving isn't it you know because the the landscape is moving so quickly that you know our clients have asked us questions and we're like okay we have to figure that out actually with you because we don't actually have an answer to some of those questions and I think the value part of agencies and again if well let me put it this way if if I'm all right I'm a client and I've got an agency and if my agency isn't coming to me and I can use GPT and mid-journing and all these other tools and I'm reasonably intelligent and I can I've got some of this up if you're not doing that at a base level and understanding what the what the the landscape looks like you've got bigger problems than you think because the clients are it's going to have a downward pressure on um on on time-based agencies anyway because if you're not adding that value piece that strategic piece and and and so forth It's a commodity game to the bottom because if you can create 100 pieces of content using the various tools and chaining in 15 20 30 at the time you did well you the client is correct to say these tools can do this now why are I getting the additional value and support or indeed should I bring this in-house now that's easy to say and hard to do because you do need you know it sounds easy but it isn't um because you do need to understand all the landscape the tools are chaining and particularly when you start doing production work the underlying weights and inferences and how the models actually work and so on as well as the IP and the copyright but it's not you know it's not rocket science so it's not it's not magic and mystery so um I think I think agencies I I think agents are ducted into it are going to have a big shot I really do because that middle management piece I mentioned and that is goes for account managers creatives performance Etc that that that little piece is quite quickly going to be production level ready it is in some cases and it's not going to stop so um if I were an agency not using this stuff I would either a partner with someone who does or B very quickly get myself a an AI policy together and get my teams leaning into it and not be careful of it because it's not going away the gas at the cat's out the box it really is so when I feel there's loads of opportunities by the way I wouldn't have started a new agency in the middle of this um fantastic um uh situation where we are if I didn't feel this was I was bullish on it and that there was massive opportunities there are and so it holds the courses right it depends how you look at it I'm an optimist a rational optimist so either this technology is a brilliant Catalyst for growth but you have to have a plan of action you have to understand it and you have to be heads into it because you're not going to outperform someone who isn't and that could be your client or your client team so you're going to have to show value anyway uh Roy we've just got a couple of minutes left um I just want to reiterate that we will be providing the recording um Allison any final questions in the last two minutes I want to be respectful of Roy's time yes so um a couple of questions that lead into each other on um reporting um so specifically um any suggestions for AI tools to do SEO reporting for clients and also for reporting in general um two clients from within the agency in order to reduce billable time do you have any suggestions for the best AI tools of reporting to clients we've got we've got we've got a a table of a thousand thumbs I'm happy to share it with you um I I I don't have one off hand um uh reason being as described I'm sure there are brilliant tools out there that other people might know um that SEO isn't isn't our key area so I haven't been spent much time in that in that space what I will say though is um tools are great but you you know you really should be thinking a little bit about your own so that might be it might be something that you know again autism ages what to do and how to prioritize their services but if you're not either a white labeling something that exists or potentially thinking about the web apps you might create or the process you might create for your clients which is IPL and ownable I mean that's that's the route I would take personally but you know I don't know maybe Jenny knows I'm not I haven't got any two-handed but I happen to share the link to the tools I'm sorry I haven't Roy okay I'm I'm going to cut it off there I think I'm just sorry I've just got one very quick last question which I think is quite important is can Google detect AI written content and if so does it affect rankings ah good great question great question yes and no I think is the fair answer um there are a load of detection tools out there which can help you to negate this and we were doing a lot of this work a few months ago that's AI driven run it through various other tools where's the percentage AI verse and it's all potential impossible so you've got to remember this as well like Google uh Google's business model is potentially a search model is potentially threatened by um openai now they've come out all guns blazing after that blip in February where they lost 100 what it was a billion a couple of billion pounds off their off their um their market cap when their AI demo went wrong so they're on their search generative experience is all going to come back into play so they've got two things they may be penalized I don't know they may be penalize um Words which aren't created with their tools don't know but the search Paradigm and the search generative experience is going to change which is the 10 Blue Links are probably going to go away because if you've got context and you've got the ability to see um where uh information is coming from and you can find it on page one we've got different paradigms so for SEO agencies I'm sure they know that um so um yes and no there was a bit who are about it uh and I'm sure that in the next Google update there'll be another piece but Google changes their mind regularly anyway and and as an SEO agency it's always difficult to get up with the eats and whatever's coming next so um my my view would be to be brutally Frank on it would be sorry for that that wasn't against the person that's the question but any content that you put out shouldn't be rubbish or shouldn't be generated without any kind of oversight anyway I'm sure they weren't doing that there wasn't the the the spirit of the question But ultimately it it you know the con if the content is good enough you're likely to be okay with Google is it valuable following the Google kind of rules that's a sensible approach definitely but that will probably change so um watch his face right thank you so much Roy for everything that you've shared um obviously when we send the follow-up we'll we'll share your contact details if you want to talk to Roy about working together um also the link to the book that Roy shared which was very generous thank you very much and the recording Alison is there anything that I've missed um no I don't think so I think Roy did mention his uh website where he'd had uh the um a suggestion of AI tools which I believe we've shared before actually on your newsletter so um we may be we'll be able to add that in as well yeah all right well thank you so much for everyone for joining us Roy thank you so much for your time Alison thank you so much for facilitating everything and um yeah enjoy enjoyed playing around with AI yourself thanks everyone Thanks for pairing with bye